Register runner when starting script

Signed-off-by: Lee Smet <lee.smet@hotmail.com>

Cargo.lock

@@ -17,6 +17,15 @@ version = "2.0.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "320119579fcad9c21884f5c4861d16174d0e06250625266f50fe6898340abefa"
 
+[[package]]
+name = "aho-corasick"
+version = "1.1.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8e60d3430d3a69478ad0993f19238d2df97c507009a52b3c10addcd7f6bcb916"
+dependencies = [
+ "memchr",
+]
+
 [[package]]
 name = "anstream"
 version = "0.6.20"
@@ -514,9 +523,10 @@ dependencies = [
  "reqwest",
  "serde",
  "serde_json",
- "thiserror 1.0.69",
+ "thiserror",
  "tokio",
  "tracing",
+ "tracing-subscriber",
 ]
 
 [[package]]
@@ -846,7 +856,7 @@ dependencies = [
  "rustc-hash",
  "serde",
  "serde_json",
- "thiserror 2.0.16",
+ "thiserror",
  "tokio",
  "tower",
  "tracing",
@@ -884,7 +894,7 @@ dependencies = [
  "serde",
  "serde_json",
  "soketto",
- "thiserror 2.0.16",
+ "thiserror",
  "tokio",
  "tokio-stream",
  "tokio-util",
@@ -901,9 +911,15 @@ dependencies = [
  "http",
  "serde",
  "serde_json",
- "thiserror 2.0.16",
+ "thiserror",
 ]
 
+[[package]]
+name = "lazy_static"
+version = "1.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "bbd2bcb4c963f2ddae06a2efc7e9f3591312473c50c6685e1f298068316e66fe"
+
 [[package]]
 name = "libc"
 version = "0.2.175"
@@ -944,6 +960,15 @@ version = "0.1.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "112b39cec0b298b6c1999fee3e31427f74f676e4cb9879ed1a121b43661a4154"
 
+[[package]]
+name = "matchers"
+version = "0.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8263075bb86c5a1b1427b5ae862e8889656f126e9f77c484496e8b47cf5c5558"
+dependencies = [
+ "regex-automata 0.1.10",
+]
+
 [[package]]
 name = "memchr"
 version = "2.7.5"
@@ -993,6 +1018,16 @@ dependencies = [
  "tempfile",
 ]
 
+[[package]]
+name = "nu-ansi-term"
+version = "0.46.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "77a8165726e8236064dbb45459242600304b42a5ea24ee2948e18e023bf7ba84"
+dependencies = [
+ "overload",
+ "winapi",
+]
+
 [[package]]
 name = "num-bigint"
 version = "0.4.6"
@@ -1086,6 +1121,12 @@ dependencies = [
  "vcpkg",
 ]
 
+[[package]]
+name = "overload"
+version = "0.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b15813163c1d831bf4a13c3610c05c0d03b39feb07f7e09fa234dac9b15aaf39"
+
 [[package]]
 name = "parking_lot"
 version = "0.12.4"
@@ -1203,7 +1244,7 @@ dependencies = [
  "rustc-hash",
  "rustls",
  "socket2 0.5.10",
- "thiserror 2.0.16",
+ "thiserror",
  "tokio",
  "tracing",
  "web-time",
@@ -1224,7 +1265,7 @@ dependencies = [
  "rustls",
  "rustls-pki-types",
  "slab",
- "thiserror 2.0.16",
+ "thiserror",
  "tinyvec",
  "tracing",
  "web-time",
@@ -1352,6 +1393,50 @@ dependencies = [
  "bitflags",
 ]
 
+[[package]]
+name = "regex"
+version = "1.11.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "23d7fd106d8c02486a8d64e778353d1cffe08ce79ac2e82f540c86d0facf6912"
+dependencies = [
+ "aho-corasick",
+ "memchr",
+ "regex-automata 0.4.10",
+ "regex-syntax 0.8.6",
+]
+
+[[package]]
+name = "regex-automata"
+version = "0.1.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6c230d73fb8d8c1b9c0b3135c5142a8acee3a0558fb8db5cf1cb65f8d7862132"
+dependencies = [
+ "regex-syntax 0.6.29",
+]
+
+[[package]]
+name = "regex-automata"
+version = "0.4.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6b9458fa0bfeeac22b5ca447c63aaf45f28439a709ccd244698632f9aa6394d6"
+dependencies = [
+ "aho-corasick",
+ "memchr",
+ "regex-syntax 0.8.6",
+]
+
+[[package]]
+name = "regex-syntax"
+version = "0.6.29"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f162c6dd7b008981e4d40210aca20b4bd0f9b60ca9271061b07f78537722f2e1"
+
+[[package]]
+name = "regex-syntax"
+version = "0.8.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "caf4aa5b0f434c91fe5c7f1ecb6a5ece2130b02ad2a590589dda5146df959001"
+
 [[package]]
 name = "reqwest"
 version = "0.12.23"
@@ -1587,6 +1672,15 @@ version = "1.0.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "bbfa15b3dddfee50a0fff136974b3e1bde555604ba463834a7eb7deb6417705d"
 
+[[package]]
+name = "sharded-slab"
+version = "0.1.7"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f40ca3c46823713e0d4209592e8d6e826aa57e928f09752619fc696c499637f6"
+dependencies = [
+ "lazy_static",
+]
+
 [[package]]
 name = "shlex"
 version = "1.3.0"
@@ -1733,33 +1827,13 @@ dependencies = [
  "windows-sys 0.60.2",
 ]
 
-[[package]]
-name = "thiserror"
-version = "1.0.69"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b6aaf5339b578ea85b50e080feb250a3e8ae8cfcdff9a461c9ec2904bc923f52"
-dependencies = [
- "thiserror-impl 1.0.69",
-]
-
 [[package]]
 name = "thiserror"
 version = "2.0.16"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "3467d614147380f2e4e374161426ff399c91084acd2363eaf549172b3d5e60c0"
 dependencies = [
- "thiserror-impl 2.0.16",
+ "thiserror-impl",
-]
-
-[[package]]
-name = "thiserror-impl"
-version = "1.0.69"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4fee6c4efc90059e10f81e6d42c60a18f76588c3d74cb83a0b242a2b6c7504c1"
-dependencies = [
- "proc-macro2",
- "quote",
- "syn",
 ]
 
 [[package]]
@@ -1773,6 +1847,15 @@ dependencies = [
  "syn",
 ]
 
+[[package]]
+name = "thread_local"
+version = "1.1.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f60246a4944f24f6e018aa17cdeffb7818b76356965d03b07d6a9886e8962185"
+dependencies = [
+ "cfg-if",
+]
+
 [[package]]
 name = "tinystr"
 version = "0.8.1"
@@ -1966,6 +2049,36 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b9d12581f227e93f094d3af2ae690a574abb8a2b9b7a96e7cfe9647b2b617678"
 dependencies = [
  "once_cell",
+ "valuable",
+]
+
+[[package]]
+name = "tracing-log"
+version = "0.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ee855f1f400bd0e5c02d150ae5de3840039a3f54b025156404e34c23c03f47c3"
+dependencies = [
+ "log",
+ "once_cell",
+ "tracing-core",
+]
+
+[[package]]
+name = "tracing-subscriber"
+version = "0.3.19"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e8189decb5ac0fa7bc8b96b7cb9b2701d60d48805aca84a238004d665fcc4008"
+dependencies = [
+ "matchers",
+ "nu-ansi-term",
+ "once_cell",
+ "regex",
+ "sharded-slab",
+ "smallvec",
+ "thread_local",
+ "tracing",
+ "tracing-core",
+ "tracing-log",
 ]
 
 [[package]]
@@ -2015,6 +2128,12 @@ version = "0.2.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "06abde3611657adf66d383f00b093d7faecc7fa57071cce2578660c9f1010821"
 
+[[package]]
+name = "valuable"
+version = "0.1.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ba73ea9cf16a25df0c8caa16c51acb937d5712a8429db78a3ee29d5dcacd3a65"
+
 [[package]]
 name = "vcpkg"
 version = "0.2.15"
@@ -2151,6 +2270,28 @@ dependencies = [
  "rustls-pki-types",
 ]
 
+[[package]]
+name = "winapi"
+version = "0.3.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5c839a674fcd7a98952e593242ea400abe93992746761e38641405d28b00f419"
+dependencies = [
+ "winapi-i686-pc-windows-gnu",
+ "winapi-x86_64-pc-windows-gnu",
+]
+
+[[package]]
+name = "winapi-i686-pc-windows-gnu"
+version = "0.4.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "ac3b87c63620426dd9b991e5ce0329eff545bccbbb34f3be09ff6fb6ab51b7b6"
+
+[[package]]
+name = "winapi-x86_64-pc-windows-gnu"
+version = "0.4.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "712e227841d057c1ee1cd2fb22fa7e5a5461ae8e48fa2ca79ec42cfc1931183f"
+
 [[package]]
 name = "windows-link"
 version = "0.1.3"

Cargo.toml

@@ -21,4 +21,5 @@ reqwest = { version = "0.12.7", features = ["json", "rustls-tls"] }
 # Base64 encoding for message payloads
 base64 = "0.22.1"
 # Error derive for clean error types
-thiserror = "1.0.64"
+thiserror = "2.0.16"
+tracing-subscriber = { version = "0.3.18", features = ["env-filter", "fmt"] }

scripts/jsonrpc_demo.py

@@ -0,0 +1,376 @@
+#!/usr/bin/env python3
+"""
+Demo script for HeroCoordinator JSON-RPC API.
+- Creates an actor
+- Verifies by loading the actor
+- Creates a context with the actor as admin/reader/executor
+- Creates three jobs with dependencies
+- Creates a flow referencing those jobs
+- Fetches and prints the flow DAG
+
+Usage:
+  COORDINATOR_URL=http://127.0.0.1:9652 python3 scripts/jsonrpc_demo.py
+Defaults to http://127.0.0.1:9652 if COORDINATOR_URL is not set.
+"""
+import os
+import json
+import sys
+from urllib import request, error
+from typing import Any, Dict, List, Tuple
+
+JSONRPC_VERSION = "2.0"
+
+class JsonRpcClient:
+    def __init__(self, url: str):
+        self.url = url.rstrip("/")
+        self._id = 0
+
+    def call(self, method: str, params: Dict[str, Any]) -> Any:
+        self._id += 1
+        payload = {
+            "jsonrpc": JSONRPC_VERSION,
+            "id": self._id,
+            "method": method,
+            "params": params,
+        }
+        data = json.dumps(payload).encode("utf-8")
+        req = request.Request(self.url, data=data, headers={"Content-Type": "application/json"})
+        try:
+            with request.urlopen(req) as resp:
+                body = resp.read()
+        except error.HTTPError as e:
+            try:
+                details = e.read().decode("utf-8", "ignore")
+            except Exception:
+                details = ""
+            raise RuntimeError(f"HTTP error {e.code}: {details}") from e
+        except error.URLError as e:
+            raise RuntimeError(f"URL error: {e.reason}") from e
+
+        try:
+            obj = json.loads(body.decode("utf-8"))
+        except Exception as e:
+            raise RuntimeError(f"Invalid JSON response: {body!r}") from e
+
+        # JSON-RPC single response expected
+        if isinstance(obj, list):
+            raise RuntimeError("Batch responses are not supported in this demo")
+
+        if obj.get("error"):
+            raise RuntimeError(f"RPC error: {json.dumps(obj['error'])}")
+
+        return obj.get("result")
+
+def print_header(title: str):
+    print("\n" + "=" * 80)
+    print(title)
+    print("=" * 80)
+
+def pretty_print(obj: Any):
+    print(json.dumps(obj, indent=2, sort_keys=True))
+
+def summarize_dag(dag: Dict[str, Any]):
+    print_header("Flow DAG Summary")
+    flow_id = dag.get("flow_id")
+    caller_id = dag.get("caller_id")
+    context_id = dag.get("context_id")
+    print(f"flow_id={flow_id} caller_id={caller_id} context_id={context_id}")
+    edges: List[Tuple[int, int]] = dag.get("edges", [])
+    roots: List[int] = dag.get("roots", [])
+    leaves: List[int] = dag.get("leaves", [])
+    levels: List[List[int]] = dag.get("levels", [])
+    nodes: Dict[str, Any] = dag.get("nodes", {})
+
+    print("Edges:")
+    for a, b in edges:
+        print(f"  {a} -> {b}")
+
+    print(f"Roots: {roots}")
+    print(f"Leaves: {leaves}")
+    print("Levels:")
+    for i, lvl in enumerate(levels):
+        print(f"  L{i}: {lvl}")
+
+    # Show nodes and their dependencies (from JobSummary)
+    print("Nodes:")
+    for k, v in nodes.items():
+        depends = v.get("depends", [])
+        prerequisites = v.get("prerequisites", [])
+        stype = v.get("script_type")
+        print(f"  Job {k}: depends={depends} prerequisites={prerequisites} script_type={stype}")
+
+def assert_edges(edges: List[Tuple[int, int]], required: List[Tuple[int, int]]):
+    edge_set = {(int(a), int(b)) for a, b in edges}
+    missing = [e for e in required if e not in edge_set]
+    if missing:
+        raise AssertionError(f"Missing expected edges in DAG: {missing}; got={sorted(edge_set)}")
+
+def main():
+    url = os.getenv("COORDINATOR_URL", "http://127.0.0.1:9652")
+    client = JsonRpcClient(url)
+
+    # Deterministic demo IDs; change if collisions happen
+    actor_id = 1001
+    context_id = 1  # Redis DB indices are 0-15; keep <= 15
+    job_a = 3001
+    job_b = 3002
+    job_c = 3003
+    job_d = 3004
+    job_e = 3005
+    job_f = 3006
+    job_g = 3007
+    job_h = 3008
+    job_i = 3009
+    flow_id = 4001
+
+runner_id = 2001
+    print_header("actor.create")
+    actor = client.call("actor.create", {
+        "actor": {
+            "id": actor_id,
+            "pubkey": "demo-pubkey",
+            "address": ["127.0.0.1"]
+        }
+    })
+    pretty_print(actor)
+
+    print_header("actor.load")
+    actor_loaded = client.call("actor.load", {"id": actor_id})
+    pretty_print(actor_loaded)
+
+    print_header("context.create")
+    context = client.call("context.create", {
+        "context": {
+            "id": context_id,
+            "admins": [actor_id],
+            "readers": [actor_id],
+            "executors": [actor_id]
+        }
+    })
+    pretty_print(context)
+print_header("runner.create")
+    runner = client.call("runner.create", {
+        "context_id": context_id,
+        "runner": {
+            "id": runner_id,
+            "pubkey": "",  # leave empty to route by IP
+            "address": "127.0.0.1",
+            "topic": f"runner{runner_id}",
+            "script_type": "Python",
+            "local": True,
+            "secret": "demo-secret"
+        }
+    })
+    pretty_print(runner)
+
+    print_header("job.create - A (root)")
+    jobA = client.call("job.create", {
+        "context_id": context_id,
+        "job": {
+            "id": job_a,
+            "caller_id": actor_id,
+            "context_id": context_id,
+            "script": "print('A')",
+            "script_type": "Python",
+            "timeout": 30,
+            "retries": 0,
+            "env_vars": {},
+            "prerequisites": [],
+            "depends": []
+        }
+    })
+    pretty_print(jobA)
+
+    print_header("job.create - B (root)")
+    jobB = client.call("job.create", {
+        "context_id": context_id,
+        "job": {
+            "id": job_b,
+            "caller_id": actor_id,
+            "context_id": context_id,
+            "script": "print('B')",
+            "script_type": "Python",
+            "timeout": 30,
+            "retries": 0,
+            "env_vars": {},
+            "prerequisites": [],
+            "depends": []
+        }
+    })
+    pretty_print(jobB)
+
+    print_header("job.create - C (depends on A and B)")
+    jobC = client.call("job.create", {
+        "context_id": context_id,
+        "job": {
+            "id": job_c,
+            "caller_id": actor_id,
+            "context_id": context_id,
+            "script": "print('C')",
+            "script_type": "Python",
+            "timeout": 30,
+            "retries": 0,
+            "env_vars": {},
+            "prerequisites": [],
+            "depends": [job_a, job_b]
+        }
+    })
+    pretty_print(jobC)
+
+    print_header("job.create - D (depends on A)")
+    jobD = client.call("job.create", {
+        "context_id": context_id,
+        "job": {
+            "id": job_d,
+            "caller_id": actor_id,
+            "context_id": context_id,
+            "script": "print('D')",
+            "script_type": "Python",
+            "timeout": 30,
+            "retries": 0,
+            "env_vars": {},
+            "prerequisites": [],
+            "depends": [job_a]
+        }
+    })
+    pretty_print(jobD)
+
+    print_header("job.create - E (depends on B)")
+    jobE = client.call("job.create", {
+        "context_id": context_id,
+        "job": {
+            "id": job_e,
+            "caller_id": actor_id,
+            "context_id": context_id,
+            "script": "print('E')",
+            "script_type": "Python",
+            "timeout": 30,
+            "retries": 0,
+            "env_vars": {},
+            "prerequisites": [],
+            "depends": [job_b]
+        }
+    })
+    pretty_print(jobE)
+
+    print_header("job.create - F (depends on C and D)")
+    jobF = client.call("job.create", {
+        "context_id": context_id,
+        "job": {
+            "id": job_f,
+            "caller_id": actor_id,
+            "context_id": context_id,
+            "script": "print('F')",
+            "script_type": "Python",
+            "timeout": 30,
+            "retries": 0,
+            "env_vars": {},
+            "prerequisites": [],
+            "depends": [job_c, job_d]
+        }
+    })
+    pretty_print(jobF)
+
+    print_header("job.create - G (depends on C and E)")
+    jobG = client.call("job.create", {
+        "context_id": context_id,
+        "job": {
+            "id": job_g,
+            "caller_id": actor_id,
+            "context_id": context_id,
+            "script": "print('G')",
+            "script_type": "Python",
+            "timeout": 30,
+            "retries": 0,
+            "env_vars": {},
+            "prerequisites": [],
+            "depends": [job_c, job_e]
+        }
+    })
+    pretty_print(jobG)
+
+    print_header("job.create - H (leaf; depends on F and G)")
+    jobH = client.call("job.create", {
+        "context_id": context_id,
+        "job": {
+            "id": job_h,
+            "caller_id": actor_id,
+            "context_id": context_id,
+            "script": "print('H')",
+            "script_type": "Python",
+            "timeout": 30,
+            "retries": 0,
+            "env_vars": {},
+            "prerequisites": [],
+            "depends": [job_f, job_g]
+        }
+    })
+    pretty_print(jobH)
+
+    print_header("job.create - I (leaf; depends on F and G)")
+    jobI = client.call("job.create", {
+        "context_id": context_id,
+        "job": {
+            "id": job_i,
+            "caller_id": actor_id,
+            "context_id": context_id,
+            "script": "print('I')",
+            "script_type": "Python",
+            "timeout": 30,
+            "retries": 0,
+            "env_vars": {},
+            "prerequisites": [],
+            "depends": [job_f, job_g]
+        }
+    })
+    pretty_print(jobI)
+
+    print_header("flow.create")
+    flow = client.call("flow.create", {
+        "context_id": context_id,
+        "flow": {
+            "id": flow_id,
+            "caller_id": actor_id,
+            "context_id": context_id,
+            "jobs": [job_a, job_b, job_c, job_d, job_e, job_f, job_g, job_h, job_i],
+            "env_vars": {}
+        }
+    })
+    pretty_print(flow)
+
+    print_header("flow.dag")
+    dag = client.call("flow.dag", {"context_id": context_id, "id": flow_id})
+    summarize_dag(dag)
+
+    # Validate roots and leaves
+    got_roots = list(map(int, dag.get("roots", [])))
+    if got_roots != sorted([job_a, job_b]):
+        print("WARNING: Unexpected roots:", got_roots, file=sys.stderr)
+
+    got_leaves = {int(x) for x in dag.get("leaves", [])}
+    expected_leaves = {job_h, job_i}
+    if got_leaves != expected_leaves:
+        print("WARNING: Unexpected leaves:", got_leaves, "expected:", expected_leaves, file=sys.stderr)
+
+    # Check edges reflect the expanded DAG
+    expected_edges = [
+        (job_a, job_c), (job_b, job_c),
+        (job_a, job_d), (job_b, job_e),
+        (job_c, job_f), (job_d, job_f),
+        (job_c, job_g), (job_e, job_g),
+        (job_f, job_h), (job_g, job_h),
+        (job_f, job_i), (job_g, job_i),
+    ]
+    try:
+        assert_edges(dag.get("edges", []), expected_edges)
+        print("DAG edges contain expected dependencies:", expected_edges)
+    except AssertionError as e:
+        print("WARNING:", e, file=sys.stderr)
+
+if __name__ == "__main__":
+    try:
+        main()
+    except Exception as e:
+        print_header("Error")
+        print(str(e))
+        sys.exit(1)

scripts/supervisor_flow_demo.py

@@ -0,0 +1,438 @@
+#!/usr/bin/env python3
+"""
+Supervisor flow demo for HeroCoordinator.
+
+This script:
+- Optionally pre-registers a Python runner on the target Supervisor over Mycelium using an admin secret (--admin-secret). If the flag is not set, this step is skipped.
+- Creates an actor
+- Creates a context granting the actor admin/reader/executor privileges
+- Registers a Runner in the context targeting a Supervisor reachable via Mycelium (by public key or IP)
+- Creates simple Python jobs (text jobs) with a small dependency chain
+- Creates a flow referencing those jobs
+- Starts the flow and polls until it finishes (or errors)
+
+Transport: JSON-RPC over HTTP to the Coordinator (default COORDINATOR_URL=http://127.0.0.1:9652).
+
+Example usage:
+  COORDINATOR_URL=http://127.0.0.1:9652 python3 scripts/supervisor_flow_demo.py --dst-ip 2001:db8::1 [--secret your-secret]
+  COORDINATOR_URL=http://127.0.0.1:9652 python3 scripts/supervisor_flow_demo.py --dst-pk bb39b4a3a4efd70f3e05e37887677e02efbda14681d0acd3882bc0f754792c32 [--secret your-secret]
+
+Notes:
+- Exactly one of --dst-ip or --dst-pk must be provided.
+- Runner.topic defaults to "supervisor.rpc" (see main.rs).
+- The router auto-discovers contexts and will deliver job.run messages to the supervisor.
+- Mycelium URL is read from MYCELIUM_URL (default http://127.0.0.1:8990).
+- supervisor.register_runner uses static name="python" and queue="python".
+"""
+
+import argparse
+import json
+import base64
+import os
+import sys
+import time
+from typing import Any, Dict, List, Optional, Tuple
+from urllib import request, error
+
+JSONRPC_VERSION = "2.0"
+
+
+def env_url() -> str:
+    return os.getenv("COORDINATOR_URL", "http://127.0.0.1:9652").rstrip("/")
+
+def env_mycelium_url() -> str:
+    return os.getenv("MYCELIUM_URL", "http://127.0.0.1:8990").rstrip("/")
+
+
+class JsonRpcClient:
+    def __init__(self, url: str):
+        self.url = url
+        self._id = 0
+
+    def call(self, method: str, params: Dict[str, Any]) -> Any:
+        self._id += 1
+        payload = {
+            "jsonrpc": JSONRPC_VERSION,
+            "id": self._id,
+            "method": method,
+            "params": params,
+        }
+        data = json.dumps(payload).encode("utf-8")
+        req = request.Request(self.url, data=data, headers={"Content-Type": "application/json"})
+        try:
+            with request.urlopen(req) as resp:
+                body = resp.read()
+        except error.HTTPError as e:
+            try:
+                details = e.read().decode("utf-8", "ignore")
+            except Exception:
+                details = ""
+            raise RuntimeError(f"HTTP error {e.code}: {details}") from e
+        except error.URLError as e:
+            raise RuntimeError(f"URL error: {e.reason}") from e
+
+        try:
+            obj = json.loads(body.decode("utf-8"))
+        except Exception as e:
+            raise RuntimeError(f"Invalid JSON response: {body!r}") from e
+
+        if isinstance(obj, list):
+            raise RuntimeError("Batch responses are not supported")
+
+        if obj.get("error"):
+            raise RuntimeError(f"RPC error: {json.dumps(obj['error'])}")
+
+        return obj.get("result")
+
+
+def print_header(title: str):
+    print("\n" + "=" * 80)
+    print(title)
+    print("=" * 80)
+
+
+def pretty(obj: Any):
+    print(json.dumps(obj, indent=2, sort_keys=True))
+
+def mycelium_register_runner(
+    myc: "JsonRpcClient",
+    dst_pk: Optional[str],
+    dst_ip: Optional[str],
+    topic: str,
+    admin_secret: str,
+    name: str = "python",
+    queue: str = "python",
+    timeout: int = 15,
+) -> Any:
+    """
+    Send supervisor.register_runner over Mycelium using pushMessage and wait for the reply.
+    - myc: JsonRpcClient for the Mycelium API (MYCELIUM_URL)
+    - dst_pk/dst_ip: destination on the overlay; one of them must be provided
+    - topic: message topic (defaults to supervisor.rpc from args)
+    - admin_secret: supervisor admin secret to authorize the registration
+    - name/queue: static identifiers for the python runner on the supervisor
+    - timeout: seconds to wait for a reply
+    Returns the JSON-RPC 'result' from the supervisor or raises on error/timeout.
+    """
+    envelope = {
+        "jsonrpc": JSONRPC_VERSION,
+        "id": 1,
+        "method": "register_runner",
+        "params": [{"secret": admin_secret, "name": name, "queue": queue}],
+    }
+    payload_b64 = base64.b64encode(json.dumps(envelope).encode("utf-8")).decode("ascii")
+    topic_b64 = base64.b64encode(topic.encode("utf-8")).decode("ascii")
+
+    if dst_pk:
+        dst = {"pk": dst_pk}
+    elif dst_ip:
+        dst = {"ip": dst_ip}
+    else:
+        raise RuntimeError("Either dst_pk or dst_ip must be provided for Mycelium destination")
+
+    params = {
+        "message": {"dst": dst, "topic": topic_b64, "payload": payload_b64},
+    }
+    resp = myc.call("pushMessage", params)
+    time.sleep(15)
+
+    # Expect an InboundMessage with a payload if a reply was received
+    # if isinstance(resp, dict) and "payload" in resp:
+    #     try:
+    #         reply = json.loads(base64.b64decode(resp["payload"]).decode("utf-8"))
+    #     except Exception as e:
+    #         raise RuntimeError(f"Invalid supervisor reply payload: {e}")
+    #     if isinstance(reply, dict) and reply.get("error"):
+    #         raise RuntimeError(f"Supervisor register_runner error: {json.dumps(reply['error'])}")
+    #     return reply.get("result")
+    #
+    # raise RuntimeError("No reply received from supervisor for register_runner (timeout)")
+
+
+def try_create_or_load(client: JsonRpcClient, create_method: str, create_params: Dict[str, Any],
+                       load_method: str, load_params: Dict[str, Any]) -> Any:
+    """Attempt a create; if it fails due to existence, try load."""
+    try:
+        return client.call(create_method, create_params)
+    except RuntimeError as e:
+        msg = str(e)
+        # Server maps AlreadyExists to StorageError, we don't have a structured error code here.
+        if "Already exists" in msg or "Storage Error" in msg or "Invalid params" in msg:
+            # Fall back to load
+            return client.call(load_method, load_params)
+        raise
+
+
+def parse_args() -> argparse.Namespace:
+    p = argparse.ArgumentParser(description="Create actor/context/runner/jobs/flow; start and wait until completion.")
+    group = p.add_mutually_exclusive_group(required=True)
+    group.add_argument("--dst-ip", help="Supervisor Mycelium IP address (IPv4 or IPv6)")
+    group.add_argument("--dst-pk", help="Supervisor public key (64-hex)")
+
+    p.add_argument("--context-id", type=int, default=2, help="Context id (Redis DB index; 0-15). Default: 2")
+    p.add_argument("--actor-id", type=int, default=11001, help="Actor id. Default: 11001")
+    p.add_argument("--runner-id", type=int, default=12001, help="Runner id. Default: 12001")
+    p.add_argument("--flow-id", type=int, default=13001, help="Flow id. Default: 13001")
+    p.add_argument("--base-job-id", type=int, default=20000, help="Base job id for first job; subsequent jobs increment. Default: 20000")
+    p.add_argument("--jobs", type=int, default=3, help="Number of jobs to create (>=1). Forms a simple chain. Default: 3")
+    p.add_argument("--timeout-secs", type=int, default=60, help="Per-job timeout seconds. Default: 60")
+    p.add_argument("--retries", type=int, default=0, help="Per-job retries (0-255). Default: 0")
+    p.add_argument(
+        "--script-type",
+        choices=["Python", "V", "Osis", "Sal"],
+        default="Python",
+        help="ScriptType for jobs/runner. Default: Python"
+    )
+    p.add_argument("--topic", default="supervisor.rpc", help="Supervisor topic. Default: supervisor.rpc")
+    p.add_argument("--secret", help="Optional supervisor secret used for authenticated supervisor calls")
+    p.add_argument("--admin-secret", help="Supervisor admin secret to pre-register a Python runner over Mycelium. If omitted, pre-registration is skipped.")
+    p.add_argument("--poll-interval", type=float, default=2.0, help="Flow poll interval seconds. Default: 2.0")
+    p.add_argument("--poll-timeout", type=int, default=600, help="Max seconds to wait for flow completion. Default: 600")
+    return p.parse_args()
+
+
+def main():
+    args = parse_args()
+    if args.jobs < 1:
+        print("ERROR: --jobs must be >= 1", file=sys.stderr)
+        sys.exit(2)
+
+    url = env_url()
+    client = JsonRpcClient(url)
+
+    mycelium_url = env_mycelium_url()
+    mycelium_client = JsonRpcClient(mycelium_url) if getattr(args, "admin_secret", None) else None
+
+    actor_id = int(args.actor_id)
+    context_id = int(args.context_id)
+    runner_id = int(args.runner_id)
+    flow_id = int(args.flow_id)
+    base_job_id = int(args.base_job_id)
+    script_type = args.script_type
+    timeout = int(args.timeout_secs)
+    retries = int(args.retries)
+    topic = args.topic
+
+    # 1) Actor
+    print_header("actor.create (or load)")
+    actor = try_create_or_load(
+        client,
+        "actor.create",
+        {
+            "actor": {
+                "id": actor_id,
+                "pubkey": "demo-pubkey",
+                "address": ["127.0.0.1"],
+            }
+        },
+        "actor.load",
+        {"id": actor_id},
+    )
+    pretty(actor)
+
+    # 2) Context
+    print_header("context.create (or load)")
+    context = try_create_or_load(
+        client,
+        "context.create",
+        {
+            "context": {
+                "id": context_id,
+                "admins": [actor_id],
+                "readers": [actor_id],
+                "executors": [actor_id],
+            }
+        },
+        "context.load",
+        {"id": context_id},
+    )
+    pretty(context)
+
+    # 3) Runner in this context
+    # Router picks pubkey if non-empty, else IP address.
+    # However, RunnerCreate requires both fields; we fill both and control routing via pubkey empty/non-empty.
+    runner_pubkey = args.dst_pk if args.dst_pk else ""
+    runner_address = args.dst_ip if args.dst_ip else "127.0.0.1"
+
+    # Optional: pre-register a Python runner on the Supervisor over Mycelium using an admin secret
+    if getattr(args, "admin_secret", None):
+        print_header("supervisor.register_runner (pre-register via Mycelium)")
+        try:
+            mycelium_result = mycelium_register_runner(
+                mycelium_client,
+                args.dst_pk if args.dst_pk else None,
+                args.dst_ip if args.dst_ip else None,
+                topic,
+                args.admin_secret,
+                name="Python",
+                queue="Python",
+                timeout=15,
+            )
+            print("Supervisor register_runner ->", mycelium_result)
+        except Exception as e:
+            print(f"ERROR: Supervisor pre-registration failed: {e}", file=sys.stderr)
+            sys.exit(1)
+
+    print_header("runner.create (or load)")
+    # runner.load requires both context_id and id
+    try:
+        runner_payload = {
+            "id": runner_id,
+            "pubkey": runner_pubkey,
+            "address": runner_address,
+            "topic": topic,
+            "script_type": script_type,
+            "local": False,
+        }
+        # Optional supervisor secret used by router for authenticated supervisor calls
+        if getattr(args, "secret", None):
+            runner_payload["secret"] = args.secret
+        runner = client.call("runner.create", {
+            "context_id": context_id,
+            "runner": runner_payload
+        })
+    except RuntimeError as e:
+        msg = str(e)
+        if "Already exists" in msg or "Storage Error" in msg or "Invalid params" in msg:
+            runner = client.call("runner.load", {"context_id": context_id, "id": runner_id})
+        else:
+            raise
+    pretty(runner)
+
+    # 4) Jobs
+    # Build a simple chain: J0 (root), J1 depends on J0, J2 depends on J1, ... up to N-1
+    job_ids: List[int] = []
+    for i in range(args.jobs):
+        jid = base_job_id + i
+        depends = [] if i == 0 else [base_job_id + (i - 1)]
+        job_payload = {
+            "id": jid,
+            "caller_id": actor_id,
+            "context_id": context_id,
+            "script": f"print('Job {i} running')",
+            "script_type": script_type,
+            "timeout": timeout,
+            "retries": retries,
+            "env_vars": {},
+            "prerequisites": [],
+            "depends": depends,
+        }
+        print_header(f"job.create - {jid} {'(root)' if not depends else f'(depends on {depends})'}")
+        try:
+            job = client.call("job.create", {
+                "context_id": context_id,
+                "job": job_payload
+            })
+        except RuntimeError as e:
+            msg = str(e)
+            if "Already exists" in msg or "Storage Error" in msg or "Invalid params" in msg:
+                job = client.call("job.load", {
+                    "context_id": context_id,
+                    "caller_id": actor_id,
+                    "id": jid
+                })
+            else:
+                raise
+        pretty(job)
+        job_ids.append(jid)
+
+    # 5) Flow
+    print_header("flow.create (or load)")
+    try:
+        flow = client.call("flow.create", {
+            "context_id": context_id,
+            "flow": {
+                "id": flow_id,
+                "caller_id": actor_id,
+                "context_id": context_id,
+                "jobs": job_ids,
+                "env_vars": {}
+            }
+        })
+    except RuntimeError as e:
+        msg = str(e)
+        if "Already exists" in msg or "Storage Error" in msg or "Invalid params" in msg:
+            flow = client.call("flow.load", {"context_id": context_id, "id": flow_id})
+        else:
+            raise
+    pretty(flow)
+
+    # Optional: show DAG
+    try:
+        print_header("flow.dag")
+        dag = client.call("flow.dag", {"context_id": context_id, "id": flow_id})
+        pretty(dag)
+    except Exception as e:
+        print(f"WARN: flow.dag failed: {e}", file=sys.stderr)
+
+    # 6) Start flow (idempotent; returns bool whether scheduler started)
+    print_header("flow.start")
+    started = client.call("flow.start", {"context_id": context_id, "id": flow_id})
+    print(f"flow.start -> {started}")
+
+    # 7) Poll until Finished or Error (or timeout)
+    print_header("Polling flow.load until completion")
+    t0 = time.time()
+    status = None
+    last_status_print = 0.0
+    poll_count = 0
+    while True:
+        poll_count += 1
+        flow = client.call("flow.load", {"context_id": context_id, "id": flow_id})
+        status = flow.get("status")
+        now = time.time()
+        if now - last_status_print >= max(1.0, float(args.poll_interval)):
+            print(f"[{int(now - t0)}s] flow.status = {status}")
+            last_status_print = now
+
+        # Every 5th poll, print the current flow DAG
+        if (poll_count % 5) == 0:
+            try:
+                print_header("flow.dag (periodic)")
+                dag = client.call("flow.dag", {"context_id": context_id, "id": flow_id})
+                pretty(dag)
+            except Exception as e:
+                print(f"WARN: periodic flow.dag failed: {e}", file=sys.stderr)
+
+        if status in ("Finished", "Error"):
+            break
+        if (now - t0) > args.poll_timeout:
+            print(f"ERROR: Flow did not complete within {args.poll_timeout}s (status={status})", file=sys.stderr)
+            break
+
+        time.sleep(float(args.poll_interval))
+
+    # 8) Final summary: job statuses
+    print_header("Final job statuses")
+    for jid in job_ids:
+        try:
+            j = client.call("job.load", {
+                "context_id": context_id,
+                "caller_id": actor_id,
+                "id": jid
+            })
+            print(f"Job {jid}: status={j.get('status')} result={j.get('result')}")
+        except Exception as e:
+            print(f"Job {jid}: load failed: {e}", file=sys.stderr)
+
+    # Exit code
+    if status == "Finished":
+        print_header("Result")
+        print("Flow finished successfully.")
+        sys.exit(0)
+    else:
+        print_header("Result")
+        print(f"Flow ended with status={status}")
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    try:
+        main()
+    except KeyboardInterrupt:
+        print("\nInterrupted.")
+        sys.exit(130)
+    except Exception as e:
+        print_header("Error")
+        print(str(e))
+        sys.exit(1)

specs/openrpc.json

@@ -563,6 +563,9 @@
           "local": {
             "type": "boolean"
           },
+          "secret": {
+            "type": "string"
+          },
           "created_at": {
             "type": "integer",
             "format": "int64"
@@ -1001,6 +1004,9 @@
           },
           "local": {
             "type": "boolean"
+          },
+          "secret": {
+            "type": "string"
           }
         }
       },

src/clients/mod.rs

@@ -1,13 +1,7 @@
-pub mod supervisor_client;
 pub mod mycelium_client;
+pub mod supervisor_client;
 pub mod types;
 
+pub use mycelium_client::{MyceliumClient, MyceliumClientError};
+pub use supervisor_client::{SupervisorClient, SupervisorClientError};
 pub use types::Destination;
-pub use supervisor_client::{
-    SupervisorClient,
-    SupervisorClientError,
-};
-pub use mycelium_client::{
-    MyceliumClient,
-    MyceliumClientError,
-};

src/clients/mycelium_client.rs

@@ -3,16 +3,18 @@ use std::sync::atomic::{AtomicU64, Ordering};
 
 use reqwest::Client as HttpClient;
 
+use base64::Engine;
+use base64::engine::general_purpose::STANDARD as BASE64_STANDARD;
 use serde_json::{Value, json};
 use thiserror::Error;
 
-use crate::models::TransportStatus;
 use crate::clients::Destination;
+use crate::models::TransportStatus;
 
 /// Lightweight client for Mycelium JSON-RPC (send + query status)
 #[derive(Clone)]
 pub struct MyceliumClient {
-    base_url: String,       // e.g. http://127.0.0.1:8990
+    base_url: String, // e.g. http://127.0.0.1:8990
     http: HttpClient,
     id_counter: Arc<AtomicU64>,
 }
@@ -53,25 +55,37 @@ impl MyceliumClient {
             "method": method,
             "params": [ params ]
         });
+
+        tracing::info!(%req, "jsonrpc");
         let resp = self.http.post(&self.base_url).json(&req).send().await?;
         let status = resp.status();
         let body: Value = resp.json().await?;
         if let Some(err) = body.get("error") {
             let code = err.get("code").and_then(|v| v.as_i64()).unwrap_or(0);
-            let msg = err.get("message").and_then(|v| v.as_str()).unwrap_or("unknown error");
+            let msg = err
+                .get("message")
+                .and_then(|v| v.as_str())
+                .unwrap_or("unknown error");
             if code == 408 {
                 return Err(MyceliumClientError::TransportTimeout);
             }
-            return Err(MyceliumClientError::RpcError(format!("code={code} msg={msg}")));
+            return Err(MyceliumClientError::RpcError(format!(
+                "code={code} msg={msg}"
+            )));
         }
         if !status.is_success() {
-            return Err(MyceliumClientError::RpcError(format!("HTTP {status}, body {body}")));
+            return Err(MyceliumClientError::RpcError(format!(
+                "HTTP {status}, body {body}"
+            )));
         }
         Ok(body)
     }
 
     /// Call messageStatus with an outbound message id (hex string)
-    pub async fn message_status(&self, id_hex: &str) -> Result<TransportStatus, MyceliumClientError> {
+    pub async fn message_status(
+        &self,
+        id_hex: &str,
+    ) -> Result<TransportStatus, MyceliumClientError> {
         let params = json!({ "id": id_hex });
         let body = self.jsonrpc("messageStatus", params).await?;
         let result = body.get("result").ok_or_else(|| {
@@ -83,7 +97,9 @@ impl MyceliumClient {
         } else if let Some(s) = result.as_str() {
             s.to_string()
         } else {
-            return Err(MyceliumClientError::InvalidResponse(format!("unexpected result shape: {result}")));
+            return Err(MyceliumClientError::InvalidResponse(format!(
+                "unexpected result shape: {result}"
+            )));
         };
         Self::map_status(&status_str).ok_or_else(|| {
             MyceliumClientError::InvalidResponse(format!("unknown status: {status_str}"))
@@ -113,16 +129,15 @@ impl MyceliumClient {
             Destination::Ip(ip) => json!({ "ip": ip.to_string() }),
             Destination::Pk(pk) => json!({ "pk": pk }),
         };
-        let message = json!({
+        let mut message = json!({
             "dst": dst_v,
             "topic": topic,
             "payload": payload_b64,
         });
-        let mut params = json!({ "message": message });
         if let Some(rt) = reply_timeout {
-            params["reply_timeout"] = json!(rt);
+            message["reply_timeout"] = json!(rt);
         }
-        params
+        message
     }
 
     /// pushMessage: send a message with dst/topic/payload. Optional reply_timeout for sync replies.
@@ -143,7 +158,87 @@ impl MyceliumClient {
 
     /// Helper to extract outbound message id from pushMessage result (InboundMessage or PushMessageResponseId)
     pub fn extract_message_id_from_result(result: &Value) -> Option<String> {
-        result.get("id").and_then(|v| v.as_str()).map(|s| s.to_string())
+        result
+            .get("id")
+            .and_then(|v| v.as_str())
+            .map(|s| s.to_string())
+    }
+    /// popMessage: retrieve an inbound message if available (optionally filtered by topic).
+    /// - peek: if true, do not remove the message from the queue
+    /// - timeout_secs: seconds to wait for a message (0 returns immediately)
+    /// - topic_plain: optional plain-text topic which will be base64-encoded per Mycelium spec
+    /// Returns:
+    ///   - Ok(Some(result_json)) on success, where result_json matches InboundMessage schema
+    ///   - Ok(None) when there is no message ready (Mycelium returns error code 204)
+    pub async fn pop_message(
+        &self,
+        peek: Option<bool>,
+        timeout_secs: Option<u64>,
+        topic_plain: Option<&str>,
+    ) -> Result<Option<Value>, MyceliumClientError> {
+        // Build params array
+        let mut params_array = vec![];
+        if let Some(p) = peek {
+            params_array.push(serde_json::Value::Bool(p));
+        } else {
+            params_array.push(serde_json::Value::Null)
+        }
+        if let Some(t) = timeout_secs {
+            params_array.push(serde_json::Value::Number(t.into()));
+        } else {
+            params_array.push(serde_json::Value::Null)
+        }
+        if let Some(tp) = topic_plain {
+            let topic_b64 = BASE64_STANDARD.encode(tp.as_bytes());
+            params_array.push(serde_json::Value::String(topic_b64));
+        } else {
+            params_array.push(serde_json::Value::Null)
+        }
+
+        let req = json!({
+            "jsonrpc": "2.0",
+            "id": self.next_id(),
+            "method": "popMessage",
+            "params": serde_json::Value::Array(params_array),
+        });
+
+        tracing::info!(%req, "calling popMessage");
+
+        let resp = self.http.post(&self.base_url).json(&req).send().await?;
+        let status = resp.status();
+        let body: Value = resp.json().await?;
+
+        // Handle JSON-RPC error envelope specially for code 204 (no message ready)
+        if let Some(err) = body.get("error") {
+            let code = err.get("code").and_then(|v| v.as_i64()).unwrap_or(0);
+            let msg = err
+                .get("message")
+                .and_then(|v| v.as_str())
+                .unwrap_or("unknown error");
+
+            if code == 204 {
+                // No message ready
+                return Ok(None);
+            }
+            if code == 408 {
+                // Align with other transport timeout mapping
+                return Err(MyceliumClientError::TransportTimeout);
+            }
+            return Err(MyceliumClientError::RpcError(format!(
+                "code={code} msg={msg}"
+            )));
+        }
+
+        if !status.is_success() {
+            return Err(MyceliumClientError::RpcError(format!(
+                "HTTP {status}, body {body}"
+            )));
+        }
+
+        let result = body.get("result").ok_or_else(|| {
+            MyceliumClientError::InvalidResponse(format!("missing result in response: {body}"))
+        })?;
+        Ok(Some(result.clone()))
     }
 }
 
@@ -162,24 +257,39 @@ mod tests {
             Some(10),
         );
         let msg1 = p1.get("message").unwrap();
-        assert_eq!(msg1.get("topic").unwrap().as_str().unwrap(), "supervisor.rpc");
+        assert_eq!(
+            msg1.get("topic").unwrap().as_str().unwrap(),
+            "supervisor.rpc"
+        );
         assert_eq!(msg1.get("payload").unwrap().as_str().unwrap(), "Zm9vYmFy");
         assert_eq!(
-            msg1.get("dst").unwrap().get("ip").unwrap().as_str().unwrap(),
+            msg1.get("dst")
+                .unwrap()
+                .get("ip")
+                .unwrap()
+                .as_str()
+                .unwrap(),
             "2001:db8::1"
         );
         assert_eq!(p1.get("reply_timeout").unwrap().as_u64().unwrap(), 10);
 
         // PK destination without timeout
         let p2 = MyceliumClient::build_push_params(
-            &Destination::Pk("bb39b4a3a4efd70f3e05e37887677e02efbda14681d0acd3882bc0f754792c32".into()),
+            &Destination::Pk(
+                "bb39b4a3a4efd70f3e05e37887677e02efbda14681d0acd3882bc0f754792c32".into(),
+            ),
             "supervisor.rpc",
             "YmF6", // "baz"
             None,
         );
         let msg2 = p2.get("message").unwrap();
         assert_eq!(
-            msg2.get("dst").unwrap().get("pk").unwrap().as_str().unwrap(),
+            msg2.get("dst")
+                .unwrap()
+                .get("pk")
+                .unwrap()
+                .as_str()
+                .unwrap(),
             "bb39b4a3a4efd70f3e05e37887677e02efbda14681d0acd3882bc0f754792c32"
         );
         assert!(p2.get("reply_timeout").is_none());

src/clients/supervisor_client.rs

@@ -99,11 +99,25 @@ impl SupervisorClient {
         })
     }
 
+    /// Build a supervisor JSON-RPC payload but force a specific id (used for correlation).
+    fn build_supervisor_payload_with_id(&self, method: &str, params: Value, id: u64) -> Value {
+        json!({
+            "jsonrpc": "2.0",
+            "id": id,
+            "method": method,
+            "params": params,
+        })
+    }
+
     fn encode_payload(payload: &Value) -> Result<String, SupervisorClientError> {
         let s = serde_json::to_string(payload)?;
         Ok(BASE64_STANDARD.encode(s.as_bytes()))
     }
 
+    fn encode_topic(topic: &[u8]) -> String {
+        BASE64_STANDARD.encode(topic)
+    }
+
     fn extract_message_id_from_result(result: &Value) -> Option<String> {
         // Two possibilities per Mycelium spec oneOf:
         // - PushMessageResponseId: { "id": "0123456789abcdef" }
@@ -120,7 +134,12 @@ impl SupervisorClient {
         let payload_b64 = Self::encode_payload(&inner)?;
         let result = self
             .mycelium
-            .push_message(&self.destination, &self.topic, &payload_b64, None)
+            .push_message(
+                &self.destination,
+                &Self::encode_topic(self.topic.as_bytes()),
+                &payload_b64,
+                None,
+            )
             .await?;
 
         if let Some(id) = MyceliumClient::extract_message_id_from_result(&result) {
@@ -138,6 +157,42 @@ impl SupervisorClient {
         )))
     }
 
+    /// Variant of call that also returns the inner supervisor JSON-RPC id used in the payload.
+    /// This id is required to correlate asynchronous popMessage replies coming from Mycelium.
+    pub async fn call_with_ids(
+        &self,
+        method: &str,
+        params: Value,
+    ) -> Result<(String, u64), SupervisorClientError> {
+        let inner_id = self.next_id();
+        let inner = self.build_supervisor_payload_with_id(method, params, inner_id);
+        let payload_b64 = Self::encode_payload(&inner)?;
+        let result = self
+            .mycelium
+            .push_message(
+                &self.destination,
+                &Self::encode_topic(self.topic.as_bytes()),
+                &payload_b64,
+                None,
+            )
+            .await?;
+
+        let out_id = if let Some(id) = MyceliumClient::extract_message_id_from_result(&result) {
+            id
+        } else if let Some(arr) = result.as_array()
+            && arr.len() == 1
+            && let Some(id) = MyceliumClient::extract_message_id_from_result(&arr[0])
+        {
+            id
+        } else {
+            return Err(SupervisorClientError::InvalidResponse(format!(
+                "result did not contain message id: {result}"
+            )));
+        };
+
+        Ok((out_id, inner_id))
+    }
+
     /// Synchronous variant: wait for a JSON-RPC reply via Mycelium reply_timeout, and return the inner JSON-RPC "result".
     /// If the supervisor returns an error object, map to RpcError.
     pub async fn call_sync(
@@ -151,7 +206,12 @@ impl SupervisorClient {
 
         let result = self
             .mycelium
-            .push_message(&self.destination, &self.topic, &payload_b64, Some(reply_timeout_secs))
+            .push_message(
+                &self.destination,
+                &Self::encode_topic(self.topic.as_bytes()),
+                &payload_b64,
+                Some(reply_timeout_secs),
+            )
             .await?;
 
         // Expect an InboundMessage-like with a base64 payload containing the supervisor JSON-RPC response
@@ -163,7 +223,11 @@ impl SupervisorClient {
                 one.get("payload")
                     .and_then(|v| v.as_str())
                     .map(|s| s.to_string())
-                    .ok_or_else(|| SupervisorClientError::InvalidResponse(format!("missing payload in result: {result}")))?
+                    .ok_or_else(|| {
+                        SupervisorClientError::InvalidResponse(format!(
+                            "missing payload in result: {result}"
+                        ))
+                    })?
             } else {
                 return Err(SupervisorClientError::TransportTimeout);
             }
@@ -174,15 +238,19 @@ impl SupervisorClient {
 
         let raw = BASE64_STANDARD
             .decode(payload_field.as_bytes())
-            .map_err(|e| SupervisorClientError::InvalidResponse(format!("invalid base64 payload: {e}")))?;
+            .map_err(|e| {
+                SupervisorClientError::InvalidResponse(format!("invalid base64 payload: {e}"))
+            })?;
         let rpc_resp: Value = serde_json::from_slice(&raw)?;
 
         if let Some(err) = rpc_resp.get("error") {
             return Err(SupervisorClientError::RpcError(err.to_string()));
         }
-        let res = rpc_resp
+        let res = rpc_resp.get("result").ok_or_else(|| {
-            .get("result")
+            SupervisorClientError::InvalidResponse(format!(
-            .ok_or_else(|| SupervisorClientError::InvalidResponse(format!("missing result in supervisor reply: {rpc_resp}")))?;
+                "missing result in supervisor reply: {rpc_resp}"
+            ))
+        })?;
         Ok(res.clone())
     }
 
@@ -286,6 +354,19 @@ impl SupervisorClient {
         self.call("job.run", params).await
     }
 
+    /// Typed wrapper returning both outbound Mycelium id and inner supervisor JSON-RPC id.
+    pub async fn job_run_with_ids(
+        &self,
+        job: Value,
+    ) -> Result<(String, u64), SupervisorClientError> {
+        let secret = self.need_secret()?;
+        let params = json!([{
+            "secret": secret,
+            "job": job
+        }]);
+        self.call_with_ids("job.run", params).await
+    }
+
     pub async fn job_start(
         &self,
         job_id: impl Into<String>,

src/lib.rs

@@ -1,8 +1,8 @@
-pub mod models;
-pub mod storage;
-pub mod service;
-mod time;
-pub mod dag;
-pub mod rpc;
 pub mod clients;
+pub mod dag;
+pub mod models;
 pub mod router;
+pub mod rpc;
+pub mod service;
+pub mod storage;
+mod time;

src/main.rs

@@ -2,6 +2,8 @@ use clap::Parser;
 use std::net::{IpAddr, SocketAddr};
 use std::sync::Arc;
 
+use tracing::{error, info};
+use tracing_subscriber::EnvFilter;
 #[derive(Debug, Clone, Parser)]
 #[command(
     name = "herocoordinator",
@@ -23,8 +25,8 @@ struct Cli {
         long = "mycelium-port",
         short = 'p',
         env = "MYCELIUM_PORT",
-        default_value_t = 9651u16,
+        default_value_t = 8990u16,
-        help = "Port for Mycelium JSON-RPC (default: 9651)"
+        help = "Port for Mycelium JSON-RPC (default: 8990)"
     )]
     mycelium_port: u16,
 
@@ -73,6 +75,14 @@ struct Cli {
 #[tokio::main]
 async fn main() {
     let cli = Cli::parse();
+    // Initialize tracing subscriber (pretty formatter; controlled by RUST_LOG)
+    let filter = EnvFilter::try_from_default_env().unwrap_or_else(|_| EnvFilter::new("info"));
+    tracing_subscriber::fmt()
+        .with_env_filter(filter)
+        .pretty()
+        .with_target(true)
+        .with_level(true)
+        .init();
 
     let http_addr = SocketAddr::new(cli.api_http_ip, cli.api_http_port);
     let ws_addr = SocketAddr::new(cli.api_ws_ip, cli.api_ws_port);
@@ -89,7 +99,7 @@ async fn main() {
     // Shared application state
     let state = Arc::new(herocoordinator::rpc::AppState::new(service));
 
-    // Start router workers (auto-discovered contexts)
+    // Start router workers (auto-discovered contexts) and a single global inbound listener
     {
         let base_url = format!("http://{}:{}", cli.mycelium_ip, cli.mycelium_port);
         let cfg = herocoordinator::router::RouterConfig {
@@ -100,6 +110,10 @@ async fn main() {
             transport_poll_interval_secs: 2,
             transport_poll_timeout_secs: 300,
         };
+        // Global inbound listener for supervisor replies via Mycelium popMessage
+        let _inbound_handle =
+            herocoordinator::router::start_inbound_listener(service_for_router.clone(), cfg.clone());
+        // Per-context outbound delivery loops
         let _auto_handle = herocoordinator::router::start_router_auto(service_for_router, cfg);
     }
 
@@ -107,10 +121,7 @@ async fn main() {
     let http_module = herocoordinator::rpc::build_module(state.clone());
     let ws_module = herocoordinator::rpc::build_module(state.clone());
 
-    println!(
+    info!(%http_addr, %ws_addr, redis_addr=%cli.redis_addr, "Starting JSON-RPC servers");
-        "Starting JSON-RPC servers: HTTP http://{} | WS ws://{} | redis_addr={}",
-        http_addr, ws_addr, cli.redis_addr
-    );
 
     // Start servers
     let _http_handle = herocoordinator::rpc::start_http(http_addr, http_module)
@@ -122,7 +133,7 @@ async fn main() {
 
     // Wait for Ctrl+C to terminate
     if let Err(e) = tokio::signal::ctrl_c().await {
-        eprintln!("Failed to listen for shutdown signal: {e}");
+        error!(error=%e, "Failed to listen for shutdown signal");
     }
-    println!("Shutdown signal received, exiting.");
+    info!("Shutdown signal received, exiting.");
 }

src/models/runner.rs

@@ -18,6 +18,8 @@ pub struct Runner {
     pub script_type: ScriptType,
     /// If this is true, the runner also listens on a local redis queue
     pub local: bool,
+    /// Optional secret used for authenticated supervisor calls (if required)
+    pub secret: Option<String>,
     pub created_at: Timestamp,
     pub updated_at: Timestamp,
 }

src/router.rs

@@ -1,13 +1,16 @@
 use std::{collections::HashSet, sync::Arc};
 
+use base64::Engine;
+use base64::engine::general_purpose::STANDARD as BASE64_STANDARD;
 use serde_json::{Value, json};
 use tokio::sync::Semaphore;
 
 use crate::{
-    clients::{Destination, SupervisorClient, MyceliumClient},
+    clients::{Destination, MyceliumClient, SupervisorClient},
     models::{Job, JobStatus, Message, MessageStatus, ScriptType, TransportStatus},
     service::AppService,
 };
+use tracing::{error, info};
 
 #[derive(Clone, Debug)]
 pub struct RouterConfig {
@@ -40,7 +43,7 @@ pub fn start_router(service: AppService, cfg: RouterConfig) -> Vec<tokio::task::
                 match MyceliumClient::new(cfg_cloned.base_url.clone()) {
                     Ok(c) => break Arc::new(c),
                     Err(e) => {
-                        eprintln!("[router ctx={}] MyceliumClient init error: {}", ctx_id, e);
+                        error!(context_id=ctx_id, error=%e, "MyceliumClient init error");
                         tokio::time::sleep(std::time::Duration::from_secs(1)).await;
                     }
                 }
@@ -70,12 +73,10 @@ pub fn start_router(service: AppService, cfg: RouterConfig) -> Vec<tokio::task::
                                 // Ensure permit is dropped at end of task
                                 let _permit = permit;
                                 if let Err(e) =
-                                    deliver_one(&service_task, &cfg_task, ctx_id, &key, mycelium).await
+                                    deliver_one(&service_task, &cfg_task, ctx_id, &key, mycelium)
+                                        .await
                                 {
-                                    eprintln!(
+                                    error!(context_id=ctx_id, key=%key, error=%e, "Delivery error");
-                                        "[router ctx={}] delivery error for {}: {}",
-                                        ctx_id, key, e
-                                    );
                                 }
                             }
                         });
@@ -85,7 +86,7 @@ pub fn start_router(service: AppService, cfg: RouterConfig) -> Vec<tokio::task::
                         continue;
                     }
                     Err(e) => {
-                        eprintln!("[router ctx={}] brpop error: {}", ctx_id, e);
+                        error!(context_id=ctx_id, error=%e, "BRPOP error");
                         // small backoff to avoid busy-loop on persistent errors
                         tokio::time::sleep(std::time::Duration::from_millis(200)).await;
                     }
@@ -141,11 +142,12 @@ async fn deliver_one(
     // Keep clones for poller usage
     let dest_for_poller = dest.clone();
     let topic_for_poller = cfg.topic.clone();
+    let secret_for_poller = runner.secret.clone();
     let client = SupervisorClient::new_with_client(
         mycelium.clone(),
         dest.clone(),
         cfg.topic.clone(),
-        None, // secret
+        runner.secret.clone(),
     );
 
     // Build supervisor method and params from Message
@@ -153,7 +155,24 @@ async fn deliver_one(
     let params = build_params(&msg)?;
 
     // Send
-    let out_id = client.call(&method, params).await?;
+    // If this is a job.run and we have a secret configured on the client,
+    // prefer the typed wrapper that injects the secret into inner supervisor params,
+    // and also capture the inner supervisor JSON-RPC id for correlation.
+    let (out_id, inner_id_opt) = if method == "job.run" {
+        if let Some(j) = msg.job.first() {
+            let jv = job_to_json(j)?;
+            // Returns (outbound message id, inner supervisor JSON-RPC id)
+            let (out, inner) = client.job_run_with_ids(jv).await?;
+            (out, Some(inner))
+        } else {
+            // Fallback: no embedded job, use the generic call
+            let out = client.call(&method, params).await?;
+            (out, None)
+        }
+    } else {
+        let out = client.call(&method, params).await?;
+        (out, None)
+    };
 
     // Store transport id and initial Sent status
     let _ = service
@@ -171,6 +190,13 @@ async fn deliver_one(
         .update_message_status(context_id, caller_id, id, MessageStatus::Acknowledged)
         .await?;
 
+    // Record correlation (inner supervisor JSON-RPC id -> job/message) for inbound popMessage handling
+    if let (Some(inner_id), Some(job_id)) = (inner_id_opt, job_id_opt) {
+        let _ = service
+            .supcorr_set(inner_id, context_id, caller_id, job_id, id)
+            .await;
+    }
+
     // Spawn transport-status poller
     {
         let service_poll = service.clone();
@@ -228,50 +254,63 @@ async fn deliver_one(
                         // Stop on terminal states
                         if matches!(s, TransportStatus::Delivered | TransportStatus::Read) {
                             // On Read, fetch supervisor job.status and update local job/message if terminal
-                            if matches!(s, TransportStatus::Read) {
+                            if matches!(s, TransportStatus::Read)
-                                if let Some(job_id) = job_id_opt {
+                                && let Some(job_id) = job_id_opt
-                                    let sup = SupervisorClient::new_with_client(
+                            {
-                                        client.clone(),
+                                let sup = SupervisorClient::new_with_client(
-                                        sup_dest.clone(),
+                                    client.clone(),
-                                        sup_topic.clone(),
+                                    sup_dest.clone(),
-                                        None,
+                                    sup_topic.clone(),
-                                    );
+                                    secret_for_poller.clone(),
-                                    match sup.job_status_sync(job_id.to_string(), 10).await {
+                                );
-                                        Ok(remote_status) => {
+                                match sup.job_status_sync(job_id.to_string(), 10).await {
-                                            if let Some((mapped, terminal)) =
+                                    Ok(remote_status) => {
-                                                map_supervisor_job_status(&remote_status)
+                                        if let Some((mapped, terminal)) =
-                                            {
+                                            map_supervisor_job_status(&remote_status)
-                                                if terminal {
+                                        {
-                                                    let _ = service_poll
+                                            if terminal {
-                                                        .update_job_status_unchecked(
+                                                let _ = service_poll
-                                                            context_id,
+                                                    .update_job_status_unchecked(
-                                                            caller_id,
+                                                        context_id,
-                                                            job_id,
+                                                        caller_id,
-                                                            mapped.clone(),
+                                                        job_id,
-                                                        )
+                                                        mapped.clone(),
-                                                        .await;
+                                                    )
+                                                    .await;
 
-                                                    // After terminal status, fetch supervisor job.result and store into Job.result
+                                                // After terminal status, fetch supervisor job.result and store into Job.result
-                                                    let sup = SupervisorClient::new_with_client(
+                                                let sup = SupervisorClient::new_with_client(
-                                                        client.clone(),
+                                                    client.clone(),
-                                                        sup_dest.clone(),
+                                                    sup_dest.clone(),
-                                                        sup_topic.clone(),
+                                                    sup_topic.clone(),
-                                                        None,
+                                                    secret_for_poller.clone(),
-                                                    );
+                                                );
-                                                    match sup.job_result_sync(job_id.to_string(), job_result_reply_timeout).await {
+                                                match sup
-                                                        Ok(result_map) => {
+                                                    .job_result_sync(
-                                                            // Persist the result into the Job.result map (merge)
+                                                        job_id.to_string(),
-                                                            let _ = service_poll
+                                                        job_result_reply_timeout,
-                                                                .update_job_result_merge_unchecked(
+                                                    )
-                                                                    context_id,
+                                                    .await
-                                                                    caller_id,
+                                                {
-                                                                    job_id,
+                                                    Ok(result_map) => {
-                                                                    result_map.clone(),
+                                                        // Persist the result into the Job.result map (merge)
-                                                                )
+                                                        let _ = service_poll
-                                                                .await;
+                                                            .update_job_result_merge_unchecked(
-                                                            // Log which key was stored (success or error)
+                                                                context_id,
-                                                            let key = result_map.keys().next().cloned().unwrap_or_else(|| "unknown".to_string());
+                                                                caller_id,
-                                                            let _ = service_poll
+                                                                job_id,
+                                                                result_map.clone(),
+                                                            )
+                                                            .await;
+                                                        // Log which key was stored (success or error)
+                                                        let key = result_map
+                                                            .keys()
+                                                            .next()
+                                                            .cloned()
+                                                            .unwrap_or_else(|| {
+                                                                "unknown".to_string()
+                                                            });
+                                                        let _ = service_poll
                                                                 .append_message_logs(
                                                                     context_id,
                                                                     caller_id,
@@ -282,9 +321,9 @@ async fn deliver_one(
                                                                     )],
                                                                 )
                                                                 .await;
-                                                        }
+                                                    }
-                                                        Err(e) => {
+                                                    Err(e) => {
-                                                            let _ = service_poll
+                                                        let _ = service_poll
                                                                 .append_message_logs(
                                                                     context_id,
                                                                     caller_id,
@@ -295,19 +334,19 @@ async fn deliver_one(
                                                                     )],
                                                                 )
                                                                 .await;
-                                                        }
                                                     }
+                                                }
 
-                                                    // Mark message as processed
+                                                // Mark message as processed
-                                                    let _ = service_poll
+                                                let _ = service_poll
-                                                        .update_message_status(
+                                                    .update_message_status(
-                                                            context_id,
+                                                        context_id,
-                                                            caller_id,
+                                                        caller_id,
-                                                            id,
+                                                        id,
-                                                            MessageStatus::Processed,
+                                                        MessageStatus::Processed,
-                                                        )
+                                                    )
-                                                        .await;
+                                                    .await;
-                                                    let _ = service_poll
+                                                let _ = service_poll
                                                         .append_message_logs(
                                                             context_id,
                                                             caller_id,
@@ -318,35 +357,31 @@ async fn deliver_one(
                                                             )],
                                                         )
                                                         .await;
-                                                }
-                                            } else {
-                                                let _ = service_poll
-                                                    .append_message_logs(
-                                                        context_id,
-                                                        caller_id,
-                                                        id,
-                                                        vec![format!(
-                                                            "Unknown supervisor status '{}' for job {}",
-                                                            remote_status, job_id
-                                                        )],
-                                                    )
-                                                    .await;
                                             }
-                                        }
+                                        } else {
-                                        Err(e) => {
                                             let _ = service_poll
                                                 .append_message_logs(
                                                     context_id,
                                                     caller_id,
                                                     id,
                                                     vec![format!(
-                                                        "job.status sync error: {}",
+                                                        "Unknown supervisor status '{}' for job {}",
-                                                        e
+                                                        remote_status, job_id
                                                     )],
                                                 )
                                                 .await;
                                         }
                                     }
+                                    Err(e) => {
+                                        let _ = service_poll
+                                            .append_message_logs(
+                                                context_id,
+                                                caller_id,
+                                                id,
+                                                vec![format!("job.status sync error: {}", e)],
+                                            )
+                                            .await;
+                                    }
                                 }
                             }
                             break;
@@ -436,7 +471,6 @@ fn map_supervisor_job_status(s: &str) -> Option<(JobStatus, bool)> {
     }
 }
 
-
 /// Auto-discover contexts periodically and ensure a router loop exists for each.
 /// Returns a JoinHandle of the discovery task (router loops are detached).
 pub fn start_router_auto(service: AppService, cfg: RouterConfig) -> tokio::task::JoinHandle<()> {
@@ -454,15 +488,169 @@ pub fn start_router_auto(service: AppService, cfg: RouterConfig) -> tokio::task:
                             };
                             let _ = start_router(service.clone(), cfg_ctx);
                             active.insert(ctx_id);
-                            eprintln!("[router] started loop for context {}", ctx_id);
+                            info!(context_id = ctx_id, "Started loop for context");
                         }
                     }
                 }
                 Err(e) => {
-                    eprintln!("[router] list_context_ids error: {}", e);
+                    error!(error=%e, "list_context_ids error");
                 }
             }
             tokio::time::sleep(std::time::Duration::from_secs(5)).await;
         }
     })
 }
+
+/// Start a single global inbound listener that reads Mycelium popMessage with topic filter,
+/// decodes supervisor JSON-RPC replies, and updates correlated jobs/messages.
+/// This listens for async replies like {"result":{"job_queued":...}} carrying the same inner JSON-RPC id.
+pub fn start_inbound_listener(
+    service: AppService,
+    cfg: RouterConfig,
+) -> tokio::task::JoinHandle<()> {
+    tokio::spawn(async move {
+        // Initialize Mycelium client (retry loop)
+        let mycelium = loop {
+            match MyceliumClient::new(cfg.base_url.clone()) {
+                Ok(c) => break c,
+                Err(e) => {
+                    error!(error=%e, "MyceliumClient init error (inbound listener)");
+                    tokio::time::sleep(std::time::Duration::from_secs(1)).await;
+                }
+            }
+        };
+
+        loop {
+            // Poll for inbound supervisor messages on the configured topic
+            match mycelium.pop_message(Some(false), Some(20), None).await {
+                Ok(Some(inb)) => {
+                    // Expect InboundMessage with base64 "payload"
+                    let Some(payload_b64) = inb.get("payload").and_then(|v| v.as_str()) else {
+                        // Not a payload-bearing message; ignore
+                        continue;
+                    };
+                    let Ok(raw) = BASE64_STANDARD.decode(payload_b64.as_bytes()) else {
+                        let _ = service
+                            .append_message_logs(
+                                0, // unknown context yet
+                                0,
+                                0,
+                                vec![
+                                    "Inbound payload base64 decode error (supervisor reply)".into(),
+                                ],
+                            )
+                            .await;
+                        continue;
+                    };
+                    tracing::info!(
+                        raw = %String::from_utf8_lossy(&raw),
+                        "Read raw messge from mycelium"
+                    );
+                    let Ok(rpc): Result<Value, _> = serde_json::from_slice(&raw) else {
+                        // Invalid JSON payload
+                        continue;
+                    };
+
+                    // Extract inner supervisor JSON-RPC id (number preferred; string fallback)
+                    let inner_id_u64 = match rpc.get("id") {
+                        Some(Value::Number(n)) => n.as_u64(),
+                        Some(Value::String(s)) => s.parse::<u64>().ok(),
+                        _ => None,
+                    };
+                    let Some(inner_id) = inner_id_u64 else {
+                        // Cannot correlate without id
+                        continue;
+                    };
+
+                    // Lookup correlation mapping
+                    match service.supcorr_get(inner_id).await {
+                        Ok(Some((context_id, caller_id, job_id, message_id))) => {
+                            // Determine success/error from supervisor JSON-RPC envelope
+                            let is_success = rpc
+                                .get("result")
+                                .map(|res| {
+                                    res.get("job_queued").is_some()
+                                        || res.as_str().map(|s| s == "job_queued").unwrap_or(false)
+                                })
+                                .unwrap_or(false);
+
+                            if is_success {
+                                // Set to Dispatched (idempotent) per spec choice, and append log
+                                let _ = service
+                                    .update_job_status_unchecked(
+                                        context_id,
+                                        caller_id,
+                                        job_id,
+                                        JobStatus::Dispatched,
+                                    )
+                                    .await;
+                                let _ = service
+                                    .append_message_logs(
+                                        context_id,
+                                        caller_id,
+                                        message_id,
+                                        vec![format!(
+                                            "Supervisor reply for job {}: job_queued",
+                                            job_id
+                                        )],
+                                    )
+                                    .await;
+                                let _ = service.supcorr_del(inner_id).await;
+                            } else if let Some(err_obj) = rpc.get("error") {
+                                // Error path: set job Error and log details
+                                let _ = service
+                                    .update_job_status_unchecked(
+                                        context_id,
+                                        caller_id,
+                                        job_id,
+                                        JobStatus::Error,
+                                    )
+                                    .await;
+                                let _ = service
+                                    .append_message_logs(
+                                        context_id,
+                                        caller_id,
+                                        message_id,
+                                        vec![format!(
+                                            "Supervisor error for job {}: {}",
+                                            job_id, err_obj
+                                        )],
+                                    )
+                                    .await;
+                                let _ = service.supcorr_del(inner_id).await;
+                            } else {
+                                // Unknown result; keep correlation for a later, clearer reply
+                                let _ = service
+                                    .append_message_logs(
+                                        context_id,
+                                        caller_id,
+                                        message_id,
+                                        vec![
+                                            "Supervisor reply did not contain job_queued or error"
+                                                .to_string(),
+                                        ],
+                                    )
+                                    .await;
+                            }
+                        }
+                        Ok(None) => {
+                            // No correlation found; ignore or log once
+                        }
+                        Err(e) => {
+                            error!(error=%e, "supcorr_get error");
+                            tokio::time::sleep(std::time::Duration::from_millis(200)).await;
+                        }
+                    }
+                }
+                Ok(None) => {
+                    // No message; continue polling
+                    continue;
+                }
+                Err(e) => {
+                    error!(error=%e, "popMessage error");
+                    tokio::time::sleep(std::time::Duration::from_millis(200)).await;
+                }
+            }
+        }
+    })
+}

src/rpc.rs

@@ -150,6 +150,8 @@ pub struct RunnerCreate {
     /// The script type this runner executes (used for routing)
     pub script_type: ScriptType,
     pub local: bool,
+    /// Optional secret used for authenticated supervisor calls (if required)
+    pub secret: Option<String>,
 }
 impl RunnerCreate {
     pub fn into_domain(self) -> Runner {
@@ -162,6 +164,7 @@ impl RunnerCreate {
             topic,
             script_type,
             local,
+            secret,
         } = self;
 
         Runner {
@@ -171,6 +174,7 @@ impl RunnerCreate {
             topic,
             script_type,
             local,
+            secret,
             created_at: ts,
             updated_at: ts,
         }
@@ -410,11 +414,7 @@ pub fn build_module(state: Arc<AppState>) -> RpcModule<()> {
                 let state = state.clone();
                 async move {
                     let p: ActorLoadParams = params.parse().map_err(invalid_params_err)?;
-                    let actor = state
+                    let actor = state.service.load_actor(p.id).await.map_err(storage_err)?;
-                        .service
-                        .load_actor(p.id)
-                        .await
-                        .map_err(storage_err)?;
                     Ok::<_, ErrorObjectOwned>(actor)
                 }
             })

src/service.rs

@@ -694,7 +694,7 @@ impl AppService {
 
         Ok(())
     }
-/// Bypass-permission variant to update a job status with transition validation.
+    /// Bypass-permission variant to update a job status with transition validation.
     /// This skips the executor permission check but enforces the same state transition rules.
     pub async fn update_job_status_unchecked(
         &self,
@@ -1161,6 +1161,37 @@ impl AppService {
     pub async fn scan_runners(&self, context_id: u32) -> Result<Vec<Runner>, BoxError> {
         self.redis.scan_runners(context_id).await
     }
+
+    /// Correlation map: store mapping from inner supervisor JSON-RPC id to context/caller/job/message.
+    pub async fn supcorr_set(
+        &self,
+        inner_id: u64,
+        context_id: u32,
+        caller_id: u32,
+        job_id: u32,
+        message_id: u32,
+    ) -> Result<(), BoxError> {
+        self.redis
+            .supcorr_set(inner_id, context_id, caller_id, job_id, message_id)
+            .await
+            .map_err(Into::into)
+    }
+
+    /// Correlation map: load mapping by inner supervisor JSON-RPC id.
+    pub async fn supcorr_get(
+        &self,
+        inner_id: u64,
+    ) -> Result<Option<(u32, u32, u32, u32)>, BoxError> {
+        self.redis
+            .supcorr_get(inner_id)
+            .await
+            .map_err(Into::into)
+    }
+
+    /// Correlation map: delete mapping by inner supervisor JSON-RPC id.
+    pub async fn supcorr_del(&self, inner_id: u64) -> Result<(), BoxError> {
+        self.redis.supcorr_del(inner_id).await.map_err(Into::into)
+    }
 }
 
 /// Auto-discovery helpers for contexts (wrappers over RedisDriver)

src/storage.rs

@@ -1,4 +1,3 @@
-
 pub mod redis;
 
 pub use redis::RedisDriver;

src/storage/redis.rs

@@ -7,8 +7,10 @@ use serde_json::{Map as JsonMap, Value};
 use tokio::sync::Mutex;
 
 use crate::models::{
-    Actor, Context, Flow, FlowStatus, Job, JobStatus, Message, MessageStatus, Runner, TransportStatus,
+    Actor, Context, Flow, FlowStatus, Job, JobStatus, Message, MessageStatus, Runner,
+    TransportStatus,
 };
+use tracing::{error, warn};
 
 type Result<T> = std::result::Result<T, Box<dyn std::error::Error + Send + Sync>>;
 
@@ -52,8 +54,14 @@ impl RedisDriver {
 
         // Slow path: create a new manager and cache it
         let url = format!("{}/{}", self.base_addr.trim_end_matches('/'), db);
-        let client = redis::Client::open(url.as_str())?;
+        let client = redis::Client::open(url.as_str()).map_err(|e| {
-        let cm = client.get_connection_manager().await?;
+            error!(%url, db=%db, error=%e, "Redis client open failed");
+            e
+        })?;
+        let cm = client.get_connection_manager().await.map_err(|e| {
+            error!(%url, db=%db, error=%e, "Redis connection manager init failed");
+            e
+        })?;
 
         let mut guard = self.managers.lock().await;
         let entry = guard.entry(db).or_insert(cm);
@@ -104,21 +112,37 @@ impl RedisDriver {
 
     async fn hset_model<T: Serialize>(&self, db: u32, key: &str, model: &T) -> Result<()> {
         let mut cm = self.manager_for_db(db).await?;
-        let pairs = Self::struct_to_hset_pairs(model)?;
+        let pairs = Self::struct_to_hset_pairs(model).map_err(|e| {
+            error!(db=%db, key=%key, error=%e, "Serialize model to HSET pairs failed");
+            e
+        })?;
         // Ensure no stale fields
-        let _: u64 = cm.del(key).await.unwrap_or(0);
+        let del_res: redis::RedisResult<u64> = cm.del(key).await;
+        if let Err(e) = del_res {
+            warn!(db=%db, key=%key, error=%e, "DEL before HSET failed");
+        }
         // Write all fields
-        let _: usize = cm.hset_multiple(key, &pairs).await?;
+        let _: () = cm.hset_multiple(key, &pairs).await.map_err(|e| {
+            error!(db=%db, key=%key, error=%e, "HSET multiple failed");
+            e
+        })?;
         Ok(())
     }
 
     async fn hget_model<T: DeserializeOwned>(&self, db: u32, key: &str) -> Result<T> {
         let mut cm = self.manager_for_db(db).await?;
-        let map: StdHashMap<String, String> = cm.hgetall(key).await?;
+        let map: StdHashMap<String, String> = cm.hgetall(key).await.map_err(|e| {
+            error!(db=%db, key=%key, error=%e, "HGETALL failed");
+            e
+        })?;
         if map.is_empty() {
+            // NotFound is expected in some flows; don't log as error
             return Err(format!("Key not found: {}", key).into());
         }
-        Self::hmap_to_struct(map)
+        Self::hmap_to_struct(map).map_err(|e| {
+            error!(db=%db, key=%key, error=%e, "Deserialize model from HGETALL failed");
+            e
+        })
     }
 
     // -----------------------------
@@ -196,7 +220,7 @@ impl RedisDriver {
         let key = Self::actor_key(id);
         self.hget_model(db, &key).await
     }
-/// Save an Actor globally in DB 0 (Actor is context-independent)
+    /// Save an Actor globally in DB 0 (Actor is context-independent)
     pub async fn save_actor_global(&self, actor: &Actor) -> Result<()> {
         let json = serde_json::to_value(actor)?;
         let id = json
@@ -299,7 +323,10 @@ impl RedisDriver {
             ("status".to_string(), status_str),
             ("updated_at".to_string(), ts.to_string()),
         ];
-        let _: usize = cm.hset_multiple(key, &pairs).await?;
+        let _: () = cm.hset_multiple(&key, &pairs).await.map_err(|e| {
+            error!(db=%db, key=%key, error=%e, "HSET update_job_status failed");
+            e
+        })?;
         Ok(())
     }
 
@@ -345,7 +372,10 @@ impl RedisDriver {
             ("status".to_string(), status_str),
             ("updated_at".to_string(), ts.to_string()),
         ];
-        let _: usize = cm.hset_multiple(key, &pairs).await?;
+        let _: () = cm.hset_multiple(&key, &pairs).await.map_err(|e| {
+            error!(db=%db, key=%key, error=%e, "HSET update_flow_status failed");
+            e
+        })?;
         Ok(())
     }
 
@@ -370,7 +400,10 @@ impl RedisDriver {
             ("status".to_string(), status_str),
             ("updated_at".to_string(), ts.to_string()),
         ];
-        let _: usize = cm.hset_multiple(key, &pairs).await?;
+        let _: () = cm.hset_multiple(&key, &pairs).await.map_err(|e| {
+            error!(db=%db, key=%key, error=%e, "HSET update_message_status failed");
+            e
+        })?;
         Ok(())
     }
 
@@ -404,7 +437,10 @@ impl RedisDriver {
         let ts = crate::time::current_timestamp();
         pairs.push(("updated_at".to_string(), ts.to_string()));
 
-        let _: usize = cm.hset_multiple(key, &pairs).await?;
+        let _: () = cm.hset_multiple(&key, &pairs).await.map_err(|e| {
+            error!(db=%db, key=%key, error=%e, "HSET update_message_transport failed");
+            e
+        })?;
         Ok(())
     }
 
@@ -437,7 +473,10 @@ impl RedisDriver {
             ("env_vars".to_string(), env_vars_str),
             ("updated_at".to_string(), ts.to_string()),
         ];
-        let _: usize = cm.hset_multiple(key, &pairs).await?;
+        let _: () = cm.hset_multiple(&key, &pairs).await.map_err(|e| {
+            error!(db=%db, key=%key, error=%e, "HSET update_flow_env_vars_merge failed");
+            e
+        })?;
         Ok(())
     }
 
@@ -470,7 +509,10 @@ impl RedisDriver {
             ("result".to_string(), result_str),
             ("updated_at".to_string(), ts.to_string()),
         ];
-        let _: usize = cm.hset_multiple(key, &pairs).await?;
+        let _: () = cm.hset_multiple(&key, &pairs).await.map_err(|e| {
+            error!(db=%db, key=%key, error=%e, "HSET update_flow_result_merge failed");
+            e
+        })?;
         Ok(())
     }
 
@@ -504,7 +546,10 @@ impl RedisDriver {
             ("env_vars".to_string(), env_vars_str),
             ("updated_at".to_string(), ts.to_string()),
         ];
-        let _: usize = cm.hset_multiple(key, &pairs).await?;
+        let _: () = cm.hset_multiple(&key, &pairs).await.map_err(|e| {
+            error!(db=%db, key=%key, error=%e, "HSET update_job_env_vars_merge failed");
+            e
+        })?;
         Ok(())
     }
 
@@ -538,7 +583,10 @@ impl RedisDriver {
             ("result".to_string(), result_str),
             ("updated_at".to_string(), ts.to_string()),
         ];
-        let _: usize = cm.hset_multiple(key, &pairs).await?;
+        let _: () = cm.hset_multiple(&key, &pairs).await.map_err(|e| {
+            error!(db=%db, key=%key, error=%e, "HSET update_job_result_merge failed");
+            e
+        })?;
         Ok(())
     }
 
@@ -553,7 +601,10 @@ impl RedisDriver {
             ("jobs".to_string(), jobs_str),
             ("updated_at".to_string(), ts.to_string()),
         ];
-        let _: usize = cm.hset_multiple(key, &pairs).await?;
+        let _: () = cm.hset_multiple(&key, &pairs).await.map_err(|e| {
+            error!(db=%db, key=%key, error=%e, "HSET update_flow_jobs_set failed");
+            e
+        })?;
         Ok(())
     }
 
@@ -584,7 +635,10 @@ impl RedisDriver {
             ("logs".to_string(), logs_str),
             ("updated_at".to_string(), ts.to_string()),
         ];
-        let _: usize = cm.hset_multiple(key, &pairs).await?;
+        let _: () = cm.hset_multiple(&key, &pairs).await.map_err(|e| {
+            error!(db=%db, key=%key, error=%e, "HSET append_message_logs failed");
+            e
+        })?;
         Ok(())
     }
 
@@ -595,7 +649,10 @@ impl RedisDriver {
     /// Push a value onto a Redis list using LPUSH in the given DB.
     pub async fn lpush_list(&self, db: u32, list: &str, value: &str) -> Result<()> {
         let mut cm = self.manager_for_db(db).await?;
-        let _: i64 = cm.lpush(list, value).await?;
+        let _: i64 = cm.lpush(list, value).await.map_err(|e| {
+            error!(db=%db, list=%list, value=%value, error=%e, "LPUSH failed");
+            e
+        })?;
         Ok(())
     }
 
@@ -615,7 +672,11 @@ impl RedisDriver {
             .arg("msg_out")
             .arg(timeout_secs)
             .query_async(&mut cm)
-            .await?;
+            .await
+            .map_err(|e| {
+                error!(db=%db, timeout_secs=%timeout_secs, error=%e, "BRPOP failed");
+                e
+            })?;
         Ok(res.map(|(_, v)| v))
     }
 
@@ -632,7 +693,11 @@ impl RedisDriver {
                 .arg("COUNT")
                 .arg(100)
                 .query_async(&mut cm)
-                .await?;
+                .await
+                .map_err(|e| {
+                    error!(db=%db, cursor=%cursor, error=%e, "SCAN failed");
+                    e
+                })?;
             for k in keys {
                 if let Ok(r) = self.hget_model::<Runner>(db, &k).await {
                     out.push(r);
@@ -653,7 +718,15 @@ impl RedisDriver {
     /// Register a context id in the global set "contexts" stored in DB 0.
     pub async fn register_context_id(&self, id: u32) -> Result<()> {
         let mut cm = self.manager_for_db(0).await?;
-        let _: i64 = redis::cmd("SADD").arg("contexts").arg(id).query_async(&mut cm).await?;
+        let _: i64 = redis::cmd("SADD")
+            .arg("contexts")
+            .arg(id)
+            .query_async(&mut cm)
+            .await
+            .map_err(|e| {
+                error!(db=0, context_id=%id, error=%e, "SADD contexts failed");
+                e
+            })?;
         Ok(())
     }
 
@@ -661,7 +734,14 @@ impl RedisDriver {
     pub async fn list_context_ids(&self) -> Result<Vec<u32>> {
         let mut cm = self.manager_for_db(0).await?;
         // Using SMEMBERS and parsing into u32
-        let vals: Vec<String> = redis::cmd("SMEMBERS").arg("contexts").query_async(&mut cm).await?;
+        let vals: Vec<String> = redis::cmd("SMEMBERS")
+            .arg("contexts")
+            .query_async(&mut cm)
+            .await
+            .map_err(|e| {
+                error!(db=0, error=%e, "SMEMBERS contexts failed");
+                e
+            })?;
         let mut out = Vec::with_capacity(vals.len());
         for v in vals {
             if let Ok(n) = v.parse::<u32>() {
@@ -671,4 +751,80 @@ impl RedisDriver {
         out.sort_unstable();
         Ok(out)
     }
+
+    // -----------------------------
+    // Supervisor correlation mapping (DB 0)
+    // Key: "supcorr:{inner_id_decimal}"
+    // Value: JSON {"context_id":u32,"caller_id":u32,"job_id":u32,"message_id":u32}
+    // TTL: 1 hour to avoid leaks in case of crashes
+    pub async fn supcorr_set(
+        &self,
+        inner_id: u64,
+        context_id: u32,
+        caller_id: u32,
+        job_id: u32,
+        message_id: u32,
+    ) -> Result<()> {
+        let mut cm = self.manager_for_db(0).await?;
+        let key = format!("supcorr:{}", inner_id);
+        let val = serde_json::json!({
+            "context_id": context_id,
+            "caller_id": caller_id,
+            "job_id": job_id,
+            "message_id": message_id,
+        })
+        .to_string();
+        // SET key val EX 3600
+        let _: () = redis::cmd("SET")
+            .arg(&key)
+            .arg(&val)
+            .arg("EX")
+            .arg(3600)
+            .query_async(&mut cm)
+            .await
+            .map_err(|e| {
+                error!(db=0, key=%key, error=%e, "SET supcorr_set failed");
+                e
+            })?;
+        Ok(())
+    }
+
+    pub async fn supcorr_get(
+        &self,
+        inner_id: u64,
+    ) -> Result<Option<(u32, u32, u32, u32)>> {
+        let mut cm = self.manager_for_db(0).await?;
+        let key = format!("supcorr:{}", inner_id);
+        let res: Option<String> = redis::cmd("GET")
+            .arg(&key)
+            .query_async(&mut cm)
+            .await
+            .map_err(|e| {
+                error!(db=0, key=%key, error=%e, "GET supcorr_get failed");
+                e
+            })?;
+        if let Some(s) = res {
+            let v: Value = serde_json::from_str(&s)?;
+            let ctx = v.get("context_id").and_then(|x| x.as_u64()).unwrap_or(0) as u32;
+            let caller = v.get("caller_id").and_then(|x| x.as_u64()).unwrap_or(0) as u32;
+            let job = v.get("job_id").and_then(|x| x.as_u64()).unwrap_or(0) as u32;
+            let msg = v.get("message_id").and_then(|x| x.as_u64()).unwrap_or(0) as u32;
+            return Ok(Some((ctx, caller, job, msg)));
+        }
+        Ok(None)
+    }
+
+    pub async fn supcorr_del(&self, inner_id: u64) -> Result<()> {
+        let mut cm = self.manager_for_db(0).await?;
+        let key = format!("supcorr:{}", inner_id);
+        let _: i64 = redis::cmd("DEL")
+            .arg(&key)
+            .query_async(&mut cm)
+            .await
+            .map_err(|e| {
+                error!(db=0, key=%key, error=%e, "DEL supcorr_del failed");
+                e
+            })?;
+        Ok(())
+    }
 }