simplicio-loop 3.32.5__tar.gz → 3.34.1__tar.gz

{simplicio_loop-3.32.5/simplicio_loop.egg-info → simplicio_loop-3.34.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: simplicio-loop
-Version: 3.32.5
+Version: 3.34.1
 Summary: The Universal Looping AI Orchestrator — a runtime-agnostic super-plugin (7 skills) that drains any queue of work end-to-end on any LLM/runtime.
 Author-email: Wesley Simplicio <wesleybob4@gmail.com>
 License: MIT
@@ -19,7 +19,7 @@ Classifier: Operating System :: OS Independent
 Requires-Python: >=3.8
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: simplicio-cli>=0.11.0
+Requires-Dist: simplicio-cli>=0.16.1
 Provides-Extra: dev
 Requires-Dist: pytest>=7; extra == "dev"
 Provides-Extra: ml
@@ -70,7 +70,7 @@ Now invoke it from your agent runtime (Claude Code, Cursor, Codex, Gemini, …):
 ## Highlights
 
 - **11 runtimes, one protocol** — Claude Code, Codex, VS Code/Copilot, Cursor, Antigravity, Kiro,
-  OpenCode, Gemini, Aider, Hermes, OpenClaw.
+  OpenCode, Gemini, Aider, Simplicio Agent (formerly Hermes), OpenClaw.
 - **Evidence-gated completion** — never a false "done"; exits only on a verified `<promise>`,
   cap, spindle handoff, or STOP.
 - **Token economy** — honest "answer concisely" baseline; savings credited only on verified-correct

{simplicio_loop-3.32.5 → simplicio_loop-3.34.1}/PYPI.md

@@ -41,7 +41,7 @@ Now invoke it from your agent runtime (Claude Code, Cursor, Codex, Gemini, …):
 ## Highlights
 
 - **11 runtimes, one protocol** — Claude Code, Codex, VS Code/Copilot, Cursor, Antigravity, Kiro,
-  OpenCode, Gemini, Aider, Hermes, OpenClaw.
+  OpenCode, Gemini, Aider, Simplicio Agent (formerly Hermes), OpenClaw.
 - **Evidence-gated completion** — never a false "done"; exits only on a verified `<promise>`,
   cap, spindle handoff, or STOP.
 - **Token economy** — honest "answer concisely" baseline; savings credited only on verified-correct

{simplicio_loop-3.32.5 → simplicio_loop-3.34.1}/README.md

@@ -10,7 +10,7 @@
   <a href="#-the-7-skills--5-accelerators"><img src="https://img.shields.io/badge/skills-7-7C3AED" alt="7 skills"></a>
   <a href="#-source-adapters"><img src="https://img.shields.io/badge/source%20adapters-5-00E08A" alt="5 source adapters"></a>
   <a href="#-11-runtimes-one-protocol"><img src="https://img.shields.io/badge/runtimes-11%20(3%20garantidos%2B8%20best--effort)-2563EB" alt="11 runtimes (3 guaranteed + 8 best-effort)"></a>
-  <a href="#-the-48-extension-points"><img src="https://img.shields.io/badge/extension%20points-48-00E08A" alt="48 extension points"></a>
+  <a href="#-the-49-extension-points"><img src="https://img.shields.io/badge/extension%20points-49-00E08A" alt="49 extension points"></a>
   <a href="#-token-economy"><img src="https://img.shields.io/badge/savings-unverified-888888" alt="Savings — unverified"></a>
   <a href="LICENSE"><img src="https://img.shields.io/badge/license-MIT-blue" alt="License"></a>
   <a href="https://discord.gg/wM6tr7xVb"><img src="https://img.shields.io/badge/Discord-Join%20Simplicio-5865F2?logo=discord&logoColor=white" alt="Join the Simplicio Discord"></a>
@@ -250,7 +250,7 @@ skill names no runtime; the runtime detects the skill.**
 | **OpenCode** | `AGENTS.md` | self-paced | MCP |
 | **Gemini** | `GEMINI.md` | self-paced | MCP / adapter |
 | **Aider** | `CONVENTIONS.md` | self-paced | — (LLM fallback) |
-| **Hermes** | native recall | native loop | **native** |
+| **Simplicio Agent** *(formerly Hermes)* | native recall | native loop | **native** |
 | **OpenClaw** | plugin SDK | native scheduler | **native** |
 
 The promise: **same protocol, same gates, same safety on all 11 — Tier 1 verified mechanically,
@@ -418,11 +418,11 @@ Two different things happen when you call **`simplicio-loop`**, and they behave
 
 | Runtime | Economy (skill) | Measurement (monitor) |
 |---|---|---|
-| **Hermes** | ✓ | ✓ **automatic** — already routed through the proxy (`base_url → :8788`) |
+| **Simplicio Agent** | ✓ | ✓ **automatic** — already routed through the proxy (`base_url → :8788`) |
 | **Claude** | ✓ (skill + hooks) | ✗ by default — Claude talks to `api.anthropic.com` directly; measured only once routed (`simplicio-cli wrap claude`, or `ANTHROPIC_BASE_URL → http://127.0.0.1:8788`) |
 | **Codex** | ✓ (skill) | ✗ by default — `simplicio-cli init codex` adds the MCP tools but does not route LLM traffic; measured with `simplicio-cli wrap codex` or an OpenAI base-url pointing at the proxy |
 
-So: the **savings happen on every runtime**; the **monitor tallies them automatically on Hermes**, and on
+So: the **savings happen on every runtime**; the **monitor tallies them automatically on Simplicio Agent**, and on
 Claude/Codex after a **one-time routing step** (`simplicio-cli wrap …` / base-url → `:8788`). Without routing,
 the economy still applies — the monitor just won't count those tokens. `scripts/simplicio-economy.sh wire`
 does this routing for OpenAI-compatible clients at install time.
@@ -432,7 +432,7 @@ does this routing for OpenAI-compatible clients at install time.
 A view of the savings you open when you want — only the capture is always-on:
 
 - **Capture proxy** — **always-on** (the one auto-started service; the wired clients need it
-  reachable). It silently captures + measures Claude + Codex + Hermes in the background.
+  reachable). It silently captures + measures Claude + Codex + Simplicio Agent in the background.
 - **Web dashboard** — `http://127.0.0.1:9090` — real-time token chart, savings gauge, the LLMs/runtimes
   and **141/144 providers (98%)** we intercept, a live proxy log. **Opens once on the first install**
   so you see it works, then it's **on-demand** — re-open it any of these ways:
@@ -530,7 +530,8 @@ cd simplicio-loop
 # install for your runtime (omit <runtime> to auto-detect)
 bash scripts/install.sh <runtime> [--global] [--minimal]        # macOS / Linux
 pwsh scripts/install.ps1 <runtime> [-Global]                    # Windows
-# <runtime> ∈ claude codex vscode cursor antigravity kiro opencode gemini aider hermes openclaw
+# <runtime> ∈ claude codex vscode cursor antigravity kiro opencode gemini aider simplicio_agent openclaw
+#            (hermes still accepted as a legacy alias for simplicio_agent)
 ```
 
 **The repo installer is full-stack by default — it installs everything.** One command sets up the whole stack:
@@ -538,7 +539,7 @@ the loop operator package (`simplicio-cli`, which exposes `simplicio-dev-cli` an
 `simplicio-mapper` transitively, auto-handling PEP 668 / externally-managed Python and symlinking
 the binaries onto `PATH`), the **full Python stack** (the package itself),
 the **7 skills + hooks** with the loop's Stop hook wired, and the **always-on capture proxy**
-with Claude + Codex + Hermes **routed and measured** in the background. The **dashboard opens once** on a
+with Claude + Codex + Simplicio Agent **routed and measured** in the background. The **dashboard opens once** on a
 fresh install, then it's on-demand (`simplicio-loop dashboard` / `simplicio-economy.sh monitor`); the
 **menu-bar tray never opens by itself** — nothing is forced to stay open.
 Pass **`--minimal`** only for headless/CI to skip the heavy deps + the machine services. Verify any time:

{simplicio_loop-3.32.5 → simplicio_loop-3.34.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "simplicio-loop"
-version = "3.32.5"
+version = "3.34.1"
 description = "The Universal Looping AI Orchestrator — a runtime-agnostic super-plugin (7 skills) that drains any queue of work end-to-end on any LLM/runtime."
 readme = "PYPI.md"
 requires-python = ">=3.8"
@@ -24,7 +24,7 @@ classifiers = [
 # The simplicio-loop drive requires the operator package (`simplicio-cli`), which exposes
 # `simplicio-dev-cli` and also brings the survey binary `simplicio-mapper` transitively.
 dependencies = [
-  "simplicio-cli>=0.11.0",
+  "simplicio-cli>=0.16.1",
 ]
 
 # Optional: the real embedding backend for `simplicio-cli semantic --ml` / `simplicio-cli rag --ml`.

{simplicio_loop-3.32.5 → simplicio_loop-3.34.1}/simplicio_loop/__init__.py

@@ -13,6 +13,6 @@ try:
     try:
         __version__ = _v("simplicio-loop")
     except PackageNotFoundError:
-        __version__ = "3.32.5"
+        __version__ = "3.34.0"
 except Exception:  # pragma: no cover
-    __version__ = "3.32.5"
+        __version__ = "3.34.0"

{simplicio_loop-3.32.5 → simplicio_loop-3.34.1}/simplicio_loop/_bundle/hooks/README.md

@@ -114,7 +114,7 @@ Add (paths relative to the repo root, or absolute):
 `orient_clamp.py`. Claude has no `afterAgentResponse`; `loop_stop.py` folds capture in by
 reading the transcript, so `loop_capture.py` isn't needed there.
 
-### Other runtimes (Codex, Gemini, Aider, OpenCode, Kiro, Antigravity, Hermes, OpenClaw)
+### Other runtimes (Codex, Gemini, Aider, OpenCode, Kiro, Antigravity, Simplicio Agent, OpenClaw)
 Most don't expose a stop hook. Use the **no-hook fallback**: the `simplicio-loop` skill
 self-paces via the host scheduler (`/loop`, OS cron, or the runtime's task scheduler), and
 `orient_clamp.py` is invoked directly. See `adapters/<runtime>/` for the per-runtime entry.

{simplicio_loop-3.32.5 → simplicio_loop-3.34.1}/simplicio_loop/_bundle/hooks/loop_stop.py

@@ -725,6 +725,22 @@ def latest_completion_receipt():
         return None
 
 
+def maintenance_deferred_receipt():
+    """Return the active maintenance-deferred completion receipt, if any.
+
+    Backlog-only maintenance is explicitly incomplete: the loop must hand off
+    with the durable receipt instead of honoring a fresh completion promise.
+    Unreadable or unrelated receipts are treated as absent.
+    """
+    try:
+        receipt = latest_completion_receipt() or {}
+        if receipt.get("ready") is False and receipt.get("reason_code") == "maintenance_deferred":
+            return receipt
+    except Exception:
+        pass
+    return None
+
+
 def completion_oracle_payload(response_text="", flow_gap=""):
     try:
         repo_root = os.getcwd()
@@ -947,23 +963,19 @@ def main():
         # Completion detection is centralized in the shared oracle (#138). The stop hook can still
         # surface watcher/flow state in the re-feed header below, but it no longer decides COMPLETE
         # on its own.
+        deferred_receipt = maintenance_deferred_receipt()
+        if deferred_receipt:
+            write_handoff("maintenance deferred (backlog-only mode active)", meta, body)
+            cleanup_and_stop()
         if promise and resp:
             oracle = completion_oracle_payload(resp, flow_gap or "")
             if oracle.get("ready") and os.path.exists(str(oracle.get("receipt_path") or "")):
                 _call_simplicio_hbp_append(iteration, promise, watcher_tag)
                 refresh_cross_agent_wiki(include_handoff=False)
                 cleanup_and_stop()  # (3) promise fulfilled → stop, no handoff needed
-            # Legacy converge fixtures and lightweight installs predate persisted run artifacts.
-            # Preserve their evidence-gated contract only when the oracle explicitly reports that
-            # no run directory exists; every persisted run remains exclusively oracle-authorized.
-            if oracle.get("reason_code") == "run_dir_missing":
-                match = PROMISE_RE.search(resp)
-                if (match and match.group(1).strip() == promise.strip()
-                        and ((not evidence_required) or has_evidence)
-                        and watcher_pass and not anchor_pending() and not flow_gap):
-                    _call_simplicio_hbp_append(iteration, promise, watcher_tag)
-                    refresh_cross_agent_wiki(include_handoff=False)
-                    cleanup_and_stop()
+            # A promise is never sufficient without a persisted run receipt.  The old
+            # lightweight fallback was a fail-open completion bypass (#138); missing run
+            # artifacts remain DELIVERY_PENDING and the loop continues or hands off at cap.
         # (3') Cursor capture may have raised the flag.
         if os.path.exists(DONE_FLAG) or os.path.exists(LEGACY_DONE_FLAG):
             oracle = completion_oracle_payload(flow_gap=flow_gap or "")

{simplicio_loop-3.32.5 → simplicio_loop-3.34.1}/simplicio_loop/_bundle/hooks/simplicio_dashboard.py

@@ -23,6 +23,7 @@ import subprocess
 import sys
 import tempfile
 import time
+import urllib.parse
 from pathlib import Path
 
 HOME = os.path.expanduser("~")
@@ -43,7 +44,8 @@ SAVINGS_JSON_CANDIDATES = [
 # Raw proxy log (Simplicio-named first; engine dir kept for back-compat).
 LOG_CANDIDATES = [
     Path(HOME) / ".simplicio" / "logs" / "proxy.log",
-    Path(HOME) / ".hermes" / "logs" / "simplicio-proxy.log",
+    Path(HOME) / ".simplicio-agent" / "logs" / "simplicio-proxy.log",
+    Path(HOME) / ".hermes" / "logs" / "simplicio-proxy.log",  # legacy alias, compat window
 ]
 LOGO_CANDIDATES = [
     REPO_ROOT / "assets" / "simplicio-loop-logo.png",
@@ -52,6 +54,9 @@ LOGO_CANDIDATES = [
 # Cross-platform temp dir (Windows has no /tmp) — must match simplicio_loop/cli.py PID_FILE.
 PID_FILE = Path(tempfile.gettempdir()) / "simplicio-token-monitor.pid"
 PROXY_PORT = os.environ.get("SIMPLICIO_PROXY_PORT", "8788")
+# Optional run projection for the visual dashboard.  It is deliberately opt-in: the token
+# monitor remains useful without a repository checkout or an active loop run.
+PROGRESS_RUN = os.environ.get("SIMPLICIO_PROGRESS_RUN", "").strip()
 # Engine call: the native Simplicio engine module, invoked cross-platform via this interpreter.
 ENGINE_CMD = [sys.executable or "python3", str(REPO_ROOT / "engine" / "simplicio_engine.py")]
 
@@ -65,7 +70,8 @@ RUNTIMES = [
     {"name": "Codex", "load": "AGENTS.md", "loop": "self-paced", "state": "partial", "intercept": "native", "logo": "openai", "proc": r"\bcodex\b", "families": ["openai"]},
     {"name": "VS Code", "load": "copilot instructions", "loop": "tasks", "state": "partial", "intercept": "native", "logo": "vscode", "proc": r"Visual Studio Code|Code Helper|Copilot", "families": ["openai"]},
     {"name": "OpenClaw", "load": "plugin SDK", "loop": "native loop", "state": "native", "intercept": "native", "logo": "openclaw", "proc": r"openclaw", "families": ["openai", "anthropic"]},
-    {"name": "Hermes", "load": "native recall", "loop": "native loop", "state": "native", "intercept": "baseurl", "logo": "hermes", "proc": r"hermes", "families": ["deepseek", "openai"]},
+    {"name": "Simplicio Agent", "load": "native recall", "loop": "native loop", "state": "native", "intercept": "baseurl", "logo": "simplicio_agent", "proc": r"simplicio-agent", "families": ["deepseek", "openai"]},
+    {"name": "Hermes (legacy)", "load": "native recall", "loop": "native loop", "state": "native", "intercept": "baseurl", "logo": "hermes", "proc": r"hermes", "families": ["deepseek", "openai"]},
     {"name": "Cursor", "load": ".cursor-plugin", "loop": "Stop hook", "state": "full", "intercept": "baseurl", "logo": "cursor", "proc": r"Cursor\.app|Cursor Helper", "families": ["openai", "anthropic"]},
     {"name": "OpenCode", "load": "AGENTS.md", "loop": "self-paced", "state": "partial", "intercept": "baseurl", "logo": "opencode", "proc": r"opencode", "families": ["openai", "anthropic"]},
     {"name": "Gemini", "load": "GEMINI.md", "loop": "self-paced", "state": "partial", "intercept": "none", "logo": "gemini", "proc": r"\bgemini\b", "families": []},
@@ -364,6 +370,11 @@ BODY = """<div class="wrap">
     </section>
   </div>
 
+  <section class="panel" id="progressPanel" hidden>
+    <div class="panel-head"><div class="title">loop progress · live receipt</div><div class="meta" id="progressMeta">waiting</div></div>
+    <div class="panel-body"><pre class="log" id="progressCard" aria-live="polite"></pre></div>
+  </section>
+
   <div class="footer">
     <span><span class="tm-g">Simplicio</span> <span class="tm-y">Token Monitor</span> · simplicio-loop</span>
     <span id="footMeta">--</span>
@@ -372,12 +383,25 @@ BODY = """<div class="wrap">
 
 SCRIPT = """<script>
 const GAUGE_CIRC = 327;
+const PROGRESS_RUN = __PROGRESS_RUN__;
+async function refreshProgress(){
+  if(!PROGRESS_RUN) return;
+  const panel=document.getElementById('progressPanel'); panel.hidden=false;
+  try{
+    const r=await fetch('/api/progress?run='+encodeURIComponent(PROGRESS_RUN),{cache:'no-store'});
+    const d=await r.json();
+    document.getElementById('progressMeta').textContent=(d.run_id||PROGRESS_RUN)+' · '+(d.status||'UNVERIFIED');
+    document.getElementById('progressCard').textContent=(d.label||'Progress')+' · '+String(d.percent??0)+'%\n'+
+      'gates: '+JSON.stringify(d.gates||{})+'\n'+(d.blockers||[]).map(x=>'blocker: '+x).join('\n');
+  }catch(e){ document.getElementById('progressMeta').textContent='UNVERIFIED · '+e; }
+}
 // Compact brand monograms for each LLM/runtime (recognizable marks, brand-tinted).
 const LOGOS = {
   claude:'<svg viewBox="0 0 24 24"><g stroke="#d97757" stroke-width="2.1" stroke-linecap="round"><path d="M12 3v18M3 12h18M5.5 5.5l13 13M18.5 5.5l-13 13"/></g></svg>',
   openai:'<svg viewBox="0 0 24 24"><g fill="none" stroke="#10a37f" stroke-width="1.8"><circle cx="12" cy="12" r="8"/><path d="M12 4v16M5 8l14 8M5 16l14-8"/></g></svg>',
   vscode:'<svg viewBox="0 0 24 24"><path fill="#3aa0e3" d="M17 2l5 2.5v15L17 22l-9-8 3-3 6 5V8l-6 5-3-3z"/><path fill="#3aa0e3" opacity=".55" d="M8 11L4 8 6 7l3 2.5z"/></svg>',
   openclaw:'<svg viewBox="0 0 24 24"><g fill="none" stroke="#ff7a3c" stroke-width="2.1" stroke-linecap="round"><path d="M6 4c0 6 1.5 11 6 14M12 4c.5 6 .3 11 0 14M18 4c0 6-1.5 11-6 14"/></g></svg>',
+  simplicio_agent:'<svg viewBox="0 0 24 24"><g stroke="#e0b341" stroke-width="2" fill="none" stroke-linecap="round"><path d="M9 5v14M15 5v14M9 12h6"/><path d="M7 8c-2.2 0-3.5 1.2-3.5 1.2M17 8c2.2 0 3.5 1.2 3.5 1.2" stroke-width="1.4"/></g></svg>',
   hermes:'<svg viewBox="0 0 24 24"><g stroke="#e0b341" stroke-width="2" fill="none" stroke-linecap="round"><path d="M9 5v14M15 5v14M9 12h6"/><path d="M7 8c-2.2 0-3.5 1.2-3.5 1.2M17 8c2.2 0 3.5 1.2 3.5 1.2" stroke-width="1.4"/></g></svg>',
   cursor:'<svg viewBox="0 0 24 24"><path fill="#cfd2d6" opacity=".28" d="M12 3l8 4.5v9L12 21l-8-4.5v-9z"/><path fill="none" stroke="#cfd2d6" stroke-width="1.5" stroke-linejoin="round" d="M12 3l8 4.5v9L12 21l-8-4.5v-9zM12 12l8-4.5M12 12v9M12 12L4 7.5"/></svg>',
   opencode:'<svg viewBox="0 0 24 24"><rect x="3" y="4" width="18" height="16" rx="2.5" fill="none" stroke="#9dff1a" stroke-width="1.6"/><path d="M7 9l3 3-3 3M13 15h4" fill="none" stroke="#9dff1a" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round"/></svg>',
@@ -558,7 +582,7 @@ async function refresh(){
     document.getElementById('stats').innerHTML=card('error',e.message,'red','',0,'red');
   }
 }
-setInterval(refresh,3000); refresh();
+setInterval(refresh,3000); refresh(); setInterval(refreshProgress,1500); refreshProgress();
 </script>"""
 
 HTML = """<!DOCTYPE html>
@@ -586,7 +610,8 @@ _FAVICON = "data:image/svg+xml;base64," + _b64.b64encode(BADGE_SVG.encode()).dec
 # Inline the badge into BODY first (re.sub does not re-scan replacement text), then
 # single-pass substitution — atomic per token, so a slot value can't expand a later slot.
 BODY = BODY.replace("__BADGE__", BADGE_SVG)
-_SLOTS = {"__FAVICON__": _FAVICON, "__STYLE__": STYLE, "__BODY__": BODY, "__SCRIPT__": SCRIPT}
+_SLOTS = {"__FAVICON__": _FAVICON, "__STYLE__": STYLE, "__BODY__": BODY,
+          "__SCRIPT__": SCRIPT.replace("__PROGRESS_RUN__", json.dumps(PROGRESS_RUN))}
 HTML = _re.sub(r"__(?:FAVICON|STYLE|BODY|SCRIPT)__", lambda m: _SLOTS[m.group(0)], HTML)
 
 
@@ -848,9 +873,56 @@ def _send_logo(handler):
     handler.wfile.write(_fallback_logo_svg().encode())
 
 
+_RUN_ID = re.compile(r"^[A-Za-z0-9][A-Za-z0-9._-]{0,127}$")
+
+
+def _progress_root():
+    configured = os.environ.get("SIMPLICIO_RUNS_DIR", "")
+    return Path(configured).expanduser() if configured else Path.cwd() / ".orchestrator" / "runs"
+
+
+def _progress_response(run_id):
+    """Return one receipt-backed progress snapshot without allowing path traversal."""
+    base = _progress_root().resolve()
+    if not run_id or not _RUN_ID.fullmatch(run_id):
+        return 400, {"schema": "simplicio.progress/v1", "status": "UNVERIFIED",
+                      "reason_code": "run_id_invalid"}
+    run_dir = (base / run_id).resolve()
+    try:
+        run_dir.relative_to(base)
+    except ValueError:
+        return 400, {"schema": "simplicio.progress/v1", "status": "UNVERIFIED",
+                      "reason_code": "run_id_outside_root"}
+    state_path = run_dir / "state.json"
+    if not state_path.is_file():
+        return 404, {"schema": "simplicio.progress/v1", "status": "UNVERIFIED",
+                     "run_id": run_id, "reason_code": "run_not_found"}
+    try:
+        from simplicio_loop.progress import build_progress
+        state = json.loads(state_path.read_text(encoding="utf-8"))
+        if not isinstance(state, dict):
+            raise ValueError("state_not_object")
+        return 200, build_progress(state, run_dir=run_dir)
+    except (OSError, ValueError, TypeError, ImportError, json.JSONDecodeError):
+        return 422, {"schema": "simplicio.progress/v1", "status": "UNVERIFIED",
+                      "run_id": run_id, "reason_code": "state_invalid"}
+
+
+def _send_json(handler, status, payload):
+    body = json.dumps(payload, ensure_ascii=False, sort_keys=True).encode("utf-8")
+    handler.send_response(status)
+    handler.send_header("Content-Type", "application/json; charset=utf-8")
+    handler.send_header("Access-Control-Allow-Origin", "*")
+    handler.send_header("Cache-Control", "no-store")
+    handler.send_header("Content-Length", str(len(body)))
+    handler.end_headers()
+    handler.wfile.write(body)
+
+
 class Handler(http.server.BaseHTTPRequestHandler):
     def do_GET(self):
-        path = self.path.split("?", 1)[0]
+        parsed = urllib.parse.urlsplit(self.path)
+        path = parsed.path
         if path == "/api/status":
             self.send_response(200)
             self.send_header("Content-Type", "application/json")
@@ -858,6 +930,11 @@ class Handler(http.server.BaseHTTPRequestHandler):
             self.send_header("Cache-Control", "no-cache")
             self.end_headers()
             self.wfile.write(json.dumps(get_status()).encode())
+        elif path == "/api/progress":
+            query = urllib.parse.parse_qs(parsed.query, keep_blank_values=True)
+            run_id = (query.get("run") or [""])[0]
+            status, payload = _progress_response(run_id)
+            _send_json(self, status, payload)
         elif path == "/assets/simplicio-logo.png":
             _send_logo(self)
         else:

{simplicio_loop-3.32.5 → simplicio_loop-3.34.1}/simplicio_loop/_bundle/scripts/cross_agent_wiki.py

@@ -223,7 +223,9 @@ def cmd_capture():
     git_log = _git_log()
     git_diff = _git_diff_stat()
     branch = _git_branch()
-    agent = os.environ.get("HERMES_PROFILE", os.environ.get("CLAUDE_PROFILE", "unknown"))
+    agent = os.environ.get("SIMPLICIO_AGENT_PROFILE",
+                           os.environ.get("HERMES_PROFILE",  # legacy alias, compat window
+                                          os.environ.get("CLAUDE_PROFILE", "unknown")))
 
     # Extract goal from scratchpad
     goal = scratchpad
@@ -526,7 +528,8 @@ def cmd_status():
 
 def cmd_selftest():
     origin = REPO
-    prior_env = os.environ.get("HERMES_PROFILE")
+    prior_env = os.environ.get("SIMPLICIO_AGENT_PROFILE")
+    prior_legacy_env = os.environ.get("HERMES_PROFILE")
     try:
         with tempfile.TemporaryDirectory() as tmp:
             _set_repo(tmp)
@@ -558,7 +561,8 @@ def cmd_selftest():
                 "status": "MEASURED",
                 "checked_at": "2026-07-01T00:00:00Z",
             }))
-            os.environ["HERMES_PROFILE"] = "selftest"
+            os.environ["SIMPLICIO_AGENT_PROFILE"] = "selftest"
+            os.environ.pop("HERMES_PROFILE", None)
 
             cmd_capture()
             cmd_summary()
@@ -577,9 +581,13 @@ def cmd_selftest():
     finally:
         _set_repo(origin)
         if prior_env is None:
+            os.environ.pop("SIMPLICIO_AGENT_PROFILE", None)
+        else:
+            os.environ["SIMPLICIO_AGENT_PROFILE"] = prior_env
+        if prior_legacy_env is None:
             os.environ.pop("HERMES_PROFILE", None)
         else:
-            os.environ["HERMES_PROFILE"] = prior_env
+            os.environ["HERMES_PROFILE"] = prior_legacy_env
 
 
 def main():

{simplicio_loop-3.32.5 → simplicio_loop-3.34.1}/simplicio_loop/_bundle/skills/simplicio-loop/SKILL.md

@@ -22,7 +22,7 @@ and safety details.
 
 ## Normative contract (non-negotiable)
 
-These invariants are MUST-level. Any runtime that loads this skill (Hermes, Claude, Cursor, or a
+These invariants are MUST-level. Any runtime that loads this skill (Simplicio Agent, Claude, Cursor, or a
 bare LLM) follows them mechanically — no paraphrase, no drift:
 
 1. **Evidence-gated exit.** The loop MUST NOT terminate without concrete evidence, produced in the
@@ -245,7 +245,7 @@ flat mode. Full phase table + usage: **`references/hierarchical-planner.md`**.
 
 Every turn's key decisions, findings, and dead-ends are captured into a persistent markdown wiki
 at `.orchestrator/wiki/` (`scripts/cross_agent_wiki.py capture|summary|handoff|status`) — a
-per-project, cross-agent, zero-friction knowledge base that survives across agent vendors (Hermes
+per-project, cross-agent, zero-friction knowledge base that survives across agent vendors (Simplicio Agent
 → Claude Code → Codex): a fresh agent reads it and sees "where we left off" with no transcript.
 Plain markdown, no vector DB. Full structure + per-turn mechanics:
 **`references/cross-agent-wiki.md`**.

simplicio_loop-3.34.1/simplicio_loop/_bundle/skills/simplicio-loop/references/control-policy.md

@@ -0,0 +1,79 @@
+# Control policy — RunProjection → LoopDecision (issue #261)
+
+`simplicio_loop/control_policy.py` is a pure convergence policy, not a second control plane. It
+takes an immutable `RunProjection` snapshot and returns a `LoopDecision`. It never opens a file,
+runs a subprocess, or keeps a parallel ledger — every input and output is a plain dict, which is
+what makes that guarantee structural rather than a convention to remember. The Runtime (or, in
+this repo, whatever driver adopts this module) remains the sole owner of queue, lease, attempt
+budget, effects, and the event log; this module only recommends a decision.
+
+## V(t) — the drift candidate
+
+```
+V = a·acs_open + b·verifiers_failed + c·effects_unverified + d·backlog + e·retry_amplification
+```
+
+Weights live in `PolicyWeights` (`a,b,c,d,e`, default `1.0, 1.0, 1.0, 0.25, 1.0`) and are published
+under `WEIGHTS_VERSION = "v1"` (`simplicio_loop.control_policy.WEIGHTS_VERSION`). Calibrate by
+passing a different `PolicyWeights` instance to `decide()` — bump `WEIGHTS_VERSION` when the
+calibrated set changes so a `LoopDecision`'s `weights_version` field stays meaningful across runs.
+
+## `LoopDecision` vocabulary
+
+`CONTINUE_SERIAL`, `CONTINUE_PARALLEL`, `OBSERVE_WAIT`, `REPLAN`, `ESCALATE`, `STOP_SUCCESS`,
+`STOP_BLOCKED`, `STOP_BUDGET`, `STOP_UNSAFE` — evaluated in that priority order inside `decide()`:
+
+1. **Hard constraints first, never traded off against V(t).** `hard_constraints.safe` /
+   `authorized` / `privacy_ok` false → `STOP_UNSAFE` (`hard_constraint_violation`).
+   `hard_constraints.within_budget` false → `STOP_BUDGET` (`budget_exhausted`). Safety, authority,
+   privacy, and budget are constraints, not weights — no drift score can outrun them.
+2. **Success**: `acs_open == 0 and verifiers_failed == 0 and effects_unverified == 0` →
+   `STOP_SUCCESS` (`verified`).
+3. **Explicit external block**: `projection.blocked` or a non-empty `blocked_reason` →
+   `STOP_BLOCKED` (reason_code = the supplied `blocked_reason`, or `external_dependency_blocked`).
+   This is the terminal state for "no drift is possible right now," distinct from a retryable stall.
+4. **Drift classification** (`classify_drift`) over `history` + the current tick:
+   - `PROGRESS` (negative delta, or no repeat yet) falls through to backpressure.
+   - `STALL` past the hysteresis `cooldown` → `REPLAN` (`stall_escalation`).
+   - `OSCILLATION` (a period-2..4 cycle of state signatures) past `cooldown` → `ESCALATE`
+     (`oscillation_detected`) — this is the case a flat fingerprint-repeat check can never see:
+     amplified retries whipsawing between two states without ever converging.
+   - Either stall/oscillation signal *under* `cooldown` → `OBSERVE_WAIT` (`hysteresis_hold`): a
+     single bad tick must never flip strategy immediately.
+5. **Backpressure** (`group_candidates`): conflict-free grouping of pending candidates by
+   read/write-set overlap, with a simple AIMD cap — a rising `capacity_signal` (errors/queue/
+   memory/IO) shrinks the max group size to 1 (multiplicative decrease); a calm signal lets it grow
+   by one (additive increase), never a fixed universal worker count. All-conflicting or empty
+   candidates → `CONTINUE_SERIAL`; otherwise `CONTINUE_PARALLEL` with the computed `groups`.
+
+## Reason-code vocabulary (reused, not reinvented)
+
+`hard_constraint_violation`, `budget_exhausted`, `verified`, `external_dependency_blocked` (or a
+caller-supplied `blocked_reason`), `hysteresis_hold`, `stall_escalation`, `oscillation_detected`,
+`no_conflict_free_parallelism`, `conflict_free_groups`. `verified` and `stall_escalation` are
+reused verbatim from `simplicio_loop/flow_semantics.py` rather than given synonyms.
+
+## Baseline traces (issue #261 Step 0)
+
+Six frozen fixtures under `tests/fixtures/control_policy/*.json`, replayed tick-by-tick by
+`tests/test_control_policy.py`: `success` (monotonic negative drift to `STOP_SUCCESS`),
+`flaky_test` (a repeat under the stall threshold that recovers), `blocked_dependency` (an explicit
+`STOP_BLOCKED`, never a silent hang), `duplicate_effect` (never claims `STOP_SUCCESS` while an
+effect is unverified), `overload` (a rising capacity signal shrinks `recommended_concurrency`), and
+`oscillation` (a period-2 cycle that reaches `ESCALATE` within the cooldown window).
+
+## Out of scope here — next steps in `simplicio-runtime`
+
+This module is the pure policy layer only. It deliberately does **not** attempt the cross-repo
+acceptance criteria from issue #261, which belong in
+[simplicio-runtime](https://github.com/wesleysimplicio/simplicio-runtime) (parent:
+simplicio-runtime#3134, related: simplicio-runtime#3042):
+
+- Migrating the Runtime to be the sole owner of queue, lease, attempt budget, and fan-out — this
+  repo's `runner.py`/`remote_queue.py` still run their own scheduling.
+- Swapping full-repo fingerprints for Mapper revision/Merkle handles.
+- Fixing the published package's Python-version support and fan-out packaging/capability manifest.
+- The ≥40% duplicate-call-reduction benchmark (or its refutation) — this requires wiring the
+  policy into a real driver loop and measuring call counts before/after.
+- Fault injection (lost event, duplicate event, stale projection, crash, flaky verifier) at the
+  Runtime's event-log/queue layer.

{simplicio_loop-3.32.5 → simplicio_loop-3.34.1}/simplicio_loop/_bundle/skills/simplicio-loop/references/cross-agent-wiki.md

@@ -5,7 +5,7 @@ Moved out of `SKILL.md` § Cross-agent persistent wiki as part of the #119 shrin
 Evolved from the one-shot `HANDOFF.md` pattern (inspired by JesseBrown1980/ai-memory).
 Every turn's key decisions, findings, and dead-ends are captured into a persistent
 markdown wiki at `.orchestrator/wiki/` — a per-project, cross-agent, zero-friction
-knowledge base that survives across agent vendors (Hermes → Claude Code → Codex).
+knowledge base that survives across agent vendors (Simplicio Agent → Claude Code → Codex).
 
 A fresh agent arriving in the repo reads the wiki and sees "where we left off"
 without needing the prior conversation transcript.

{simplicio_loop-3.32.5/simplicio_loop/_bundle/skills/simplicio-tasks → simplicio_loop-3.34.1/simplicio_loop/_bundle/skills/simplicio-loop}/references/extension-points.md

@@ -1,4 +1,4 @@
-# Extension points — the 48 named binding points
+# Extension points — the 49 named binding points
 
 These are the named points where work happens. For each, if the host runtime exposes a native
 capability, BIND it (deterministic, local-first, near-zero token). If not, the LLM performs the
@@ -29,6 +29,7 @@ fallback with standard tools. The skill depends on the ABSTRACTION, never on a s
 | `human_gate` | Async human approval channel | ask user inline |
 | `shell_exec` | Clamped shell execution (structured output, bounded size) | Bash with `\| head -N` |
 | `retry` | Classified retry+backoff by failure class | manual retry loop |
+| `convergence_policy` | Pure `RunProjection → LoopDecision` control policy (issue #261): a Lyapunov-style drift candidate V(t) over open ACs/failed verifiers/unverified effects/backlog/retry-amplification, versioned weights, stall **and** period-2+ oscillation detection, hysteresis before REPLAN/ESCALATE, and a conflict-free backpressure/parallelism recommendation (AIMD-style, no fixed worker count). Runnable form: `simplicio_loop/control_policy.py` (`decide()`) — zero I/O, zero subprocess, zero ledger; the Runtime remains sole owner of queue/lease/attempt-budget and applies the recommendation, it never obeys blindly (see `control-policy.md`). | LLM reasons about retry/parallelism by hand from the run journal and current AC/verifier counts |
 | `status` | Live observability dashboard | `gh` queries |
 | `security` | Supply-chain / secret scan | `rg` for secrets |
 | `intake` | Ingest work from sprint/board link — or, with no board, the local backlog (`task_backlog.py next`) | `gh issue list` |

{simplicio_loop-3.32.5 → simplicio_loop-3.34.1}/simplicio_loop/_bundle/skills/simplicio-loop/references/token-capture.md

@@ -16,7 +16,7 @@ or install the engine's durable integration for X. Some runtimes can do neither.
 
 ## Providers we intercept (the real breadth)
 
-Interception is really about **providers**, not apps. From the Hermes/OpenCode provider lists
+Interception is really about **providers**, not apps. From the Simplicio Agent/OpenCode provider lists
 (`app/providers.json`, derived from `~/.hermes/models_dev_cache.json` + the active providers),
 **141 of 144 providers (98%)** are interceptable — every OpenAI-compatible endpoint (139) plus
 Anthropic (2). Only 3 are Google-native (`google`, `vertex`, `gemini`) and need their own shim.
@@ -44,7 +44,7 @@ elsewhere, with a headless print fallback.
 | Tier | Runtimes | How capture is wired | Notes |
 |---|---|---|---|
 | **native** | Claude · Codex · VS Code (Copilot) · OpenClaw | `simplicio-cli capture init <client>` (engine's durable hooks + **transparent** provider routing) | Forwards to the client's REAL provider — does not change the model. |
-| **base-url** | Hermes · Cursor · OpenCode | point the client's model `base_url` at `http://127.0.0.1:<port>` (OpenAI/Anthropic-compatible) | Hermes is wired this way today (→ DeepSeek). |
+| **base-url** | Simplicio Agent · Cursor · OpenCode | point the client's model `base_url` at `http://127.0.0.1:<port>` (OpenAI/Anthropic-compatible) | Simplicio Agent is wired this way today (→ DeepSeek). |
 | **none** | Gemini · Kiro · Antigravity | — | Proprietary Google/AWS APIs the proxy can't speak; not interceptable yet. |
 
 **7 of 10 runtimes are interceptable; 3 are not.** The dashboard shows this live with per-runtime
@@ -53,7 +53,7 @@ logos and tier badges, and dims the non-interceptable ones.
 ## Critical caveat — proxy backend
 
 A single proxy forwards to **one upstream**. The proxy installed by `setup_simplicio.sh` targets
-DeepSeek (for Hermes). If you route a *different* client (e.g. Codex) through that same proxy, its
+DeepSeek (for Simplicio Agent). If you route a *different* client (e.g. Codex) through that same proxy, its
 requests also go to DeepSeek — that's a **model swap, not transparent capture**.
 
 For real multi-runtime capture you want the engine's **transparent provider routing** (each client
@@ -96,12 +96,12 @@ bash scripts/simplicio-economy.sh capture anthropic
 ```
 
 `status` reports: capture proxy, token monitor (`:9090`), menu-bar tray, the deterministic operator
-`simplicio-dev-cli`, the **auto-capture wiring** (Claude · Codex/OpenAI · Hermes), and lifetime savings.
+`simplicio-dev-cli`, the **auto-capture wiring** (Claude · Codex/OpenAI · Simplicio Agent), and lifetime savings.
 
 ### `wire` — all three runtimes measured, automatically
 
-`wire` (run by `setup_simplicio.sh` at the end of install) makes the monitor measure **Hermes +
-Claude + Codex** with no manual step. It sets, in the shell profile (cross-platform via
+`wire` (run by `setup_simplicio.sh` at the end of install) makes the monitor measure **Simplicio
+Agent + Claude + Codex** with no manual step. It sets, in the shell profile (cross-platform via
 `install_services.py wire` — `setx` on Windows):
 
 - `OPENAI_BASE_URL = http://127.0.0.1:<port>/v1` → Codex / Cursor / OpenCode / any OpenAI client
@@ -121,4 +121,4 @@ A **transparent** capture proxy (`--openai-api-url https://api.openai.com/v1`) w
 gpt-5.4-2026-03-05`, `usage` 108+4 tokens) and the proxy's own `/stats` recorded
 `api_requests: 4`, `primary_model: gpt-4o-mini`, `total_tokens_before: 124`. Proof that routing an
 OpenAI client through the transparent proxy **captures its tokens without changing its model** — the
-correct way to capture Codex/Cursor/OpenCode, kept separate from the Hermes→DeepSeek proxy.
+correct way to capture Codex/Cursor/OpenCode, kept separate from the Simplicio Agent→DeepSeek proxy.

{simplicio_loop-3.32.5/simplicio_loop/_bundle/skills/simplicio-tasks → simplicio_loop-3.34.1/simplicio_loop/_bundle/skills/simplicio-loop}/references/token-economy.md

@@ -141,7 +141,7 @@ Detect platform once: `python3 -c "import platform; print(platform.system())"`
 | Comment / close | `gh issue comment N --body "…"` · `gh issue close N` |
 | Workflow dispatch | `gh api repos/{owner}/{repo}/dispatches -f event_type=X` |
 
-### node (vscode, openclaw, claude-code) / python (hermes)
+### node (vscode, openclaw, claude-code) / python (simplicio-agent, formerly hermes)
 | intent | command |
 |---|---|
 | Deterministic install | `npm ci` · `pnpm install --frozen-lockfile` · `uv sync --locked` |
@@ -160,7 +160,7 @@ Detect platform once: `python3 -c "import platform; print(platform.system())"`
 | Azure login / acct | `az login` · `az account show -o json` |
 | Azure DevOps boards (if used) | `az boards work-item show --id N` · `az boards query --wiql "…"` · `az repos pr list` · `az pipelines runs list` |
 
-> NOTE: the 5 scanned local repos (hermes-agent, openclaw, vscode, codex, claude-code) use
+> NOTE: the 5 scanned local repos (simplicio-agent (formerly hermes-agent), openclaw, vscode, codex, claude-code) use
 > **GitHub (`gh`) + GitHub Actions exclusively** — no Azure DevOps. The `az`/`az boards` rows are
 > provided for repos that DO use Azure DevOps; bind them as a `source_adapter` only when detected.
 

{simplicio_loop-3.32.5 → simplicio_loop-3.34.1}/simplicio_loop/_bundle/skills/simplicio-loop/references/web-evidence.md

@@ -34,7 +34,7 @@ npx playwright install --with-deps chromium
 PWTEST_OUTPUT_DIR=.orchestrator/tee/web npx playwright test --trace on --output .orchestrator/tee/web
 ```
 
-**Fallback B — playwright-python / pytest (Python repos, e.g. hermes-agent):**
+**Fallback B — playwright-python / pytest (Python repos, e.g. the Simplicio Agent repo, formerly hermes-agent):**
 ```bash
 pip install playwright pytest-playwright && playwright install chromium
 pytest --tracing retain-on-failure --output .orchestrator/tee/web

{simplicio_loop-3.32.5 → simplicio_loop-3.34.1}/simplicio_loop/_bundle/skills/simplicio-tasks/SKILL.md

@@ -17,7 +17,7 @@ mention `/simplicio-tasks` keep working without breaking.
 
 | Need depth on… | Read |
 |---|---|
-| the 48 extension points + fallbacks | `references/extension-points.md` |
+| the 49 extension points + fallbacks | `references/extension-points.md` |
 | token economy (catalog, caps, clamp, tee+CCR, terminal table) | `references/token-economy.md` |
 | discover / intake / route / autoscale / speed / model-routing | `references/orchestration.md` |
 | quality loop · safety gates · delivery · feedback | `references/quality-safety-delivery.md` |