agentorch 2.2.0__tar.gz → 2.3.0__tar.gz

{agentorch-2.2.0 → agentorch-2.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agentorch
-Version: 2.2.0
+Version: 2.3.0
 Summary: DevTorch — AI reasoning capture, audit trail, and governance for agent-assisted development
 Author-email: Hemant Joshi <hemant@flotorch.ai>
 License: Apache-2.0
@@ -54,10 +54,10 @@ Requires-Dist: agentorch[cloud,embeddings,mcp,proxy,wrapper]; extra == "all"
 
 **Local-first AI agent governance.** DevTorch gives multi-agent systems a shared on-disk audit trail, coordination vector, privacy controls, and LLM integration layer — all without a remote service.
 
-> **Latest release:** `agentorch 2.2.0` — Reasoning Plus Learning: semantic relevance, learning deduplication, provenance tracking, LLM-based extraction, and a feedback loop. Install or upgrade with `pip install -U agentorch`.
+> **Latest release:** `agentorch 2.3.0` — Always-on Capture & Daemon: background file watcher, git-native event capture, universal LLM proxy routes, and `devtorch daemon`. Install or upgrade with `pip install -U agentorch`.
 
 ```bash
-pip install -U agentorch>=2.1.0
+pip install -U agentorch>=2.3.0
 devtorch init
 devtorch setup          # connects to Claude Code, Cursor, or Antigravity
 devtorch debug timeline
@@ -130,7 +130,7 @@ DevTorch preserves your `.GCC/` store across upgrades. Newer releases add IDE co
 **Upgrade any time:**
 
 ```bash
-pip install -U agentorch>=2.1.0
+pip install -U agentorch>=2.3.0
 cd /path/to/your/project
 devtorch setup              # or: devtorch setup --target <ide>
 devtorch doctor
@@ -194,7 +194,7 @@ Open Claude Code in your project. On the first task Claude will call `devtorch_c
 
 **What you get:** Full RACP integration — reasoning commits, sensitivity events, coordination vector Θ, divergence detection, HITL conflict resolution.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch setup` will rewrite `.claude/hooks.json`, `.claude/settings.json`, and `CLAUDE.md` with the latest versions.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch setup` will rewrite `.claude/hooks.json`, `.claude/settings.json`, and `CLAUDE.md` with the latest versions.
 
 ---
 
@@ -238,7 +238,7 @@ devtorch doctor
 
 **What you get:** MCP tools, VS Code sidebar, optional HTTP proxy for passive capture of all LLM calls (`devtorch proxy start`).
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch setup --target cursor` will regenerate `.cursor/mcp.json` and update `CLAUDE.md`.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch setup --target cursor` will regenerate `.cursor/mcp.json` and update `CLAUDE.md`.
 
 ---
 
@@ -270,7 +270,7 @@ devtorch doctor
 
 **What you get:** Full RACP integration. The multi-agent nature of Antigravity makes Θ accumulation especially valuable — multiple agents share the same reasoning state through `devtorch_context`.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch setup --target antigravity` will regenerate the MCP config and update `AGENTS.md`.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch setup --target antigravity` will regenerate the MCP config and update `AGENTS.md`.
 
 ---
 
@@ -308,7 +308,7 @@ Open the project in OpenCode. The DevTorch MCP tools will be listed, and the mod
 
 **What you get:** Cloud-hosted MCP server integration with automatic local fallback, automatic reasoning commits via `AGENTS.md` instructions, and a shared Θ state across sessions.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch cloud client-config opencode --org <org> --repo <repo> --api-key <api-key>` will regenerate `opencode.json` with the latest bridge and fallback settings. **Restart OpenCode** after updating the config.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch cloud client-config opencode --org <org> --repo <repo> --api-key <api-key>` will regenerate `opencode.json` with the latest bridge and fallback settings. **Restart OpenCode** after updating the config.
 
 ---
 
@@ -340,7 +340,7 @@ devtorch sensitivity add --source you --concept auth --confidence 0.9 --disclosu
 
 **What you get:** Sidebar panel (read-only), git-commit-level event capture, manual reasoning commits. Full capture requires switching to Claude Code, Cursor, Kiro, or Antigravity.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch setup` will reinstall the git hook and update the governance instructions.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch setup` will reinstall the git hook and update the governance instructions.
 
 ---
 
@@ -405,7 +405,7 @@ devtorch doctor
 
 **What you get:** Full RACP integration equivalent to Claude Code — MCP tools, steering-document governance, agent action hooks.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0`, then re-run the manual steps above and refresh your steering document with the latest `AGENTS.md` text.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0`, then re-run the manual steps above and refresh your steering document with the latest `AGENTS.md` text.
 
 ---
 
@@ -438,7 +438,7 @@ All `invoke_model()` and `converse()` calls are identical. DevTorch injects the
 devtorch doctor
 ```
 
-> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.1.0`.
+> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.3.0`.
 
 ---
 
@@ -472,7 +472,7 @@ client = AzureOpenAI(
 devtorch doctor
 ```
 
-> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.1.0`.
+> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.3.0`.
 
 ---
 
@@ -498,7 +498,7 @@ codex "Add error handling to the payment service"
 
 All Codex calls are intercepted by the proxy, which injects the RACP prefix and captures responses into `.GCC/`.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0`.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0`.
 
 ---
 
@@ -532,7 +532,7 @@ devtorch doctor
 
 No API key required. Fully local — RACP context and captures stay on-device.
 
-> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.1.0`.
+> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.3.0`.
 
 ---
 
@@ -598,7 +598,7 @@ Supported `<ide>` values: `opencode`, `claude`, `cursor`, `vscode`, `stdio`.
 
 > **Updating from an older version:** You do **not** need to redeploy the Cloudflare Worker for client-side changes. Just regenerate the IDE config on each machine after upgrading `agentorch`:
 > ```bash
-> > pip install -U agentorch>=2.1.0
+> > pip install -U agentorch>=2.3.0
 > > export DEVTORCH_MCP_DOMAIN=devtorch-mcp.YOUR_ACCOUNT.workers.dev
 > > devtorch cloud client-config <ide> --org <org> --repo <repo> --api-key <api-key>
 > > ```

{agentorch-2.2.0 → agentorch-2.3.0}/README.md

@@ -2,10 +2,10 @@
 
 **Local-first AI agent governance.** DevTorch gives multi-agent systems a shared on-disk audit trail, coordination vector, privacy controls, and LLM integration layer — all without a remote service.
 
-> **Latest release:** `agentorch 2.2.0` — Reasoning Plus Learning: semantic relevance, learning deduplication, provenance tracking, LLM-based extraction, and a feedback loop. Install or upgrade with `pip install -U agentorch`.
+> **Latest release:** `agentorch 2.3.0` — Always-on Capture & Daemon: background file watcher, git-native event capture, universal LLM proxy routes, and `devtorch daemon`. Install or upgrade with `pip install -U agentorch`.
 
 ```bash
-pip install -U agentorch>=2.1.0
+pip install -U agentorch>=2.3.0
 devtorch init
 devtorch setup          # connects to Claude Code, Cursor, or Antigravity
 devtorch debug timeline
@@ -78,7 +78,7 @@ DevTorch preserves your `.GCC/` store across upgrades. Newer releases add IDE co
 **Upgrade any time:**
 
 ```bash
-pip install -U agentorch>=2.1.0
+pip install -U agentorch>=2.3.0
 cd /path/to/your/project
 devtorch setup              # or: devtorch setup --target <ide>
 devtorch doctor
@@ -142,7 +142,7 @@ Open Claude Code in your project. On the first task Claude will call `devtorch_c
 
 **What you get:** Full RACP integration — reasoning commits, sensitivity events, coordination vector Θ, divergence detection, HITL conflict resolution.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch setup` will rewrite `.claude/hooks.json`, `.claude/settings.json`, and `CLAUDE.md` with the latest versions.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch setup` will rewrite `.claude/hooks.json`, `.claude/settings.json`, and `CLAUDE.md` with the latest versions.
 
 ---
 
@@ -186,7 +186,7 @@ devtorch doctor
 
 **What you get:** MCP tools, VS Code sidebar, optional HTTP proxy for passive capture of all LLM calls (`devtorch proxy start`).
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch setup --target cursor` will regenerate `.cursor/mcp.json` and update `CLAUDE.md`.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch setup --target cursor` will regenerate `.cursor/mcp.json` and update `CLAUDE.md`.
 
 ---
 
@@ -218,7 +218,7 @@ devtorch doctor
 
 **What you get:** Full RACP integration. The multi-agent nature of Antigravity makes Θ accumulation especially valuable — multiple agents share the same reasoning state through `devtorch_context`.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch setup --target antigravity` will regenerate the MCP config and update `AGENTS.md`.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch setup --target antigravity` will regenerate the MCP config and update `AGENTS.md`.
 
 ---
 
@@ -256,7 +256,7 @@ Open the project in OpenCode. The DevTorch MCP tools will be listed, and the mod
 
 **What you get:** Cloud-hosted MCP server integration with automatic local fallback, automatic reasoning commits via `AGENTS.md` instructions, and a shared Θ state across sessions.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch cloud client-config opencode --org <org> --repo <repo> --api-key <api-key>` will regenerate `opencode.json` with the latest bridge and fallback settings. **Restart OpenCode** after updating the config.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch cloud client-config opencode --org <org> --repo <repo> --api-key <api-key>` will regenerate `opencode.json` with the latest bridge and fallback settings. **Restart OpenCode** after updating the config.
 
 ---
 
@@ -288,7 +288,7 @@ devtorch sensitivity add --source you --concept auth --confidence 0.9 --disclosu
 
 **What you get:** Sidebar panel (read-only), git-commit-level event capture, manual reasoning commits. Full capture requires switching to Claude Code, Cursor, Kiro, or Antigravity.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch setup` will reinstall the git hook and update the governance instructions.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch setup` will reinstall the git hook and update the governance instructions.
 
 ---
 
@@ -353,7 +353,7 @@ devtorch doctor
 
 **What you get:** Full RACP integration equivalent to Claude Code — MCP tools, steering-document governance, agent action hooks.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0`, then re-run the manual steps above and refresh your steering document with the latest `AGENTS.md` text.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0`, then re-run the manual steps above and refresh your steering document with the latest `AGENTS.md` text.
 
 ---
 
@@ -386,7 +386,7 @@ All `invoke_model()` and `converse()` calls are identical. DevTorch injects the
 devtorch doctor
 ```
 
-> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.1.0`.
+> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.3.0`.
 
 ---
 
@@ -420,7 +420,7 @@ client = AzureOpenAI(
 devtorch doctor
 ```
 
-> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.1.0`.
+> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.3.0`.
 
 ---
 
@@ -446,7 +446,7 @@ codex "Add error handling to the payment service"
 
 All Codex calls are intercepted by the proxy, which injects the RACP prefix and captures responses into `.GCC/`.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0`.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0`.
 
 ---
 
@@ -480,7 +480,7 @@ devtorch doctor
 
 No API key required. Fully local — RACP context and captures stay on-device.
 
-> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.1.0`.
+> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.3.0`.
 
 ---
 
@@ -546,7 +546,7 @@ Supported `<ide>` values: `opencode`, `claude`, `cursor`, `vscode`, `stdio`.
 
 > **Updating from an older version:** You do **not** need to redeploy the Cloudflare Worker for client-side changes. Just regenerate the IDE config on each machine after upgrading `agentorch`:
 > ```bash
-> > pip install -U agentorch>=2.1.0
+> > pip install -U agentorch>=2.3.0
 > > export DEVTORCH_MCP_DOMAIN=devtorch-mcp.YOUR_ACCOUNT.workers.dev
 > > devtorch cloud client-config <ide> --org <org> --repo <repo> --api-key <api-key>
 > > ```

{agentorch-2.2.0 → agentorch-2.3.0}/agentorch.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agentorch
-Version: 2.2.0
+Version: 2.3.0
 Summary: DevTorch — AI reasoning capture, audit trail, and governance for agent-assisted development
 Author-email: Hemant Joshi <hemant@flotorch.ai>
 License: Apache-2.0
@@ -54,10 +54,10 @@ Requires-Dist: agentorch[cloud,embeddings,mcp,proxy,wrapper]; extra == "all"
 
 **Local-first AI agent governance.** DevTorch gives multi-agent systems a shared on-disk audit trail, coordination vector, privacy controls, and LLM integration layer — all without a remote service.
 
-> **Latest release:** `agentorch 2.2.0` — Reasoning Plus Learning: semantic relevance, learning deduplication, provenance tracking, LLM-based extraction, and a feedback loop. Install or upgrade with `pip install -U agentorch`.
+> **Latest release:** `agentorch 2.3.0` — Always-on Capture & Daemon: background file watcher, git-native event capture, universal LLM proxy routes, and `devtorch daemon`. Install or upgrade with `pip install -U agentorch`.
 
 ```bash
-pip install -U agentorch>=2.1.0
+pip install -U agentorch>=2.3.0
 devtorch init
 devtorch setup          # connects to Claude Code, Cursor, or Antigravity
 devtorch debug timeline
@@ -130,7 +130,7 @@ DevTorch preserves your `.GCC/` store across upgrades. Newer releases add IDE co
 **Upgrade any time:**
 
 ```bash
-pip install -U agentorch>=2.1.0
+pip install -U agentorch>=2.3.0
 cd /path/to/your/project
 devtorch setup              # or: devtorch setup --target <ide>
 devtorch doctor
@@ -194,7 +194,7 @@ Open Claude Code in your project. On the first task Claude will call `devtorch_c
 
 **What you get:** Full RACP integration — reasoning commits, sensitivity events, coordination vector Θ, divergence detection, HITL conflict resolution.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch setup` will rewrite `.claude/hooks.json`, `.claude/settings.json`, and `CLAUDE.md` with the latest versions.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch setup` will rewrite `.claude/hooks.json`, `.claude/settings.json`, and `CLAUDE.md` with the latest versions.
 
 ---
 
@@ -238,7 +238,7 @@ devtorch doctor
 
 **What you get:** MCP tools, VS Code sidebar, optional HTTP proxy for passive capture of all LLM calls (`devtorch proxy start`).
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch setup --target cursor` will regenerate `.cursor/mcp.json` and update `CLAUDE.md`.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch setup --target cursor` will regenerate `.cursor/mcp.json` and update `CLAUDE.md`.
 
 ---
 
@@ -270,7 +270,7 @@ devtorch doctor
 
 **What you get:** Full RACP integration. The multi-agent nature of Antigravity makes Θ accumulation especially valuable — multiple agents share the same reasoning state through `devtorch_context`.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch setup --target antigravity` will regenerate the MCP config and update `AGENTS.md`.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch setup --target antigravity` will regenerate the MCP config and update `AGENTS.md`.
 
 ---
 
@@ -308,7 +308,7 @@ Open the project in OpenCode. The DevTorch MCP tools will be listed, and the mod
 
 **What you get:** Cloud-hosted MCP server integration with automatic local fallback, automatic reasoning commits via `AGENTS.md` instructions, and a shared Θ state across sessions.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch cloud client-config opencode --org <org> --repo <repo> --api-key <api-key>` will regenerate `opencode.json` with the latest bridge and fallback settings. **Restart OpenCode** after updating the config.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch cloud client-config opencode --org <org> --repo <repo> --api-key <api-key>` will regenerate `opencode.json` with the latest bridge and fallback settings. **Restart OpenCode** after updating the config.
 
 ---
 
@@ -340,7 +340,7 @@ devtorch sensitivity add --source you --concept auth --confidence 0.9 --disclosu
 
 **What you get:** Sidebar panel (read-only), git-commit-level event capture, manual reasoning commits. Full capture requires switching to Claude Code, Cursor, Kiro, or Antigravity.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0 && devtorch setup` will reinstall the git hook and update the governance instructions.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0 && devtorch setup` will reinstall the git hook and update the governance instructions.
 
 ---
 
@@ -405,7 +405,7 @@ devtorch doctor
 
 **What you get:** Full RACP integration equivalent to Claude Code — MCP tools, steering-document governance, agent action hooks.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0`, then re-run the manual steps above and refresh your steering document with the latest `AGENTS.md` text.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0`, then re-run the manual steps above and refresh your steering document with the latest `AGENTS.md` text.
 
 ---
 
@@ -438,7 +438,7 @@ All `invoke_model()` and `converse()` calls are identical. DevTorch injects the
 devtorch doctor
 ```
 
-> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.1.0`.
+> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.3.0`.
 
 ---
 
@@ -472,7 +472,7 @@ client = AzureOpenAI(
 devtorch doctor
 ```
 
-> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.1.0`.
+> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.3.0`.
 
 ---
 
@@ -498,7 +498,7 @@ codex "Add error handling to the payment service"
 
 All Codex calls are intercepted by the proxy, which injects the RACP prefix and captures responses into `.GCC/`.
 
-> **Updating from an older version:** `pip install -U agentorch>=2.1.0`.
+> **Updating from an older version:** `pip install -U agentorch>=2.3.0`.
 
 ---
 
@@ -532,7 +532,7 @@ devtorch doctor
 
 No API key required. Fully local — RACP context and captures stay on-device.
 
-> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.1.0`.
+> **Updating from an older version:** `pip install -U agentorch[wrapper]>=2.3.0`.
 
 ---
 
@@ -598,7 +598,7 @@ Supported `<ide>` values: `opencode`, `claude`, `cursor`, `vscode`, `stdio`.
 
 > **Updating from an older version:** You do **not** need to redeploy the Cloudflare Worker for client-side changes. Just regenerate the IDE config on each machine after upgrading `agentorch`:
 > ```bash
-> > pip install -U agentorch>=2.1.0
+> > pip install -U agentorch>=2.3.0
 > > export DEVTORCH_MCP_DOMAIN=devtorch-mcp.YOUR_ACCOUNT.workers.dev
 > > devtorch cloud client-config <ide> --org <org> --repo <repo> --api-key <api-key>
 > > ```

{agentorch-2.2.0 → agentorch-2.3.0}/agentorch.egg-info/SOURCES.txt

@@ -68,6 +68,9 @@ devtorch_core/codex/proxy.py
 devtorch_core/consolidation/__init__.py
 devtorch_core/consolidation/synthesizer.py
 devtorch_core/consolidation/workflow.py
+devtorch_core/daemon/__init__.py
+devtorch_core/daemon/supervisor.py
+devtorch_core/daemon/watcher.py
 devtorch_core/divergence/__init__.py
 devtorch_core/divergence/detector.py
 devtorch_core/gateway/__init__.py
@@ -87,6 +90,7 @@ devtorch_core/hitl/channels.py
 devtorch_core/hitl/orchestrator.py
 devtorch_core/hooks/__init__.py
 devtorch_core/hooks/claude_code.py
+devtorch_core/hooks/git_capture.py
 devtorch_core/hooks/git_commit.py
 devtorch_core/hooks/installer.py
 devtorch_core/hooks/runner.py
@@ -123,7 +127,10 @@ devtorch_core/proxy/__init__.py
 devtorch_core/proxy/server.py
 devtorch_core/proxy/routes/__init__.py
 devtorch_core/proxy/routes/anthropic.py
+devtorch_core/proxy/routes/azure_openai.py
 devtorch_core/proxy/routes/gemini.py
+devtorch_core/proxy/routes/groq.py
+devtorch_core/proxy/routes/ollama.py
 devtorch_core/proxy/routes/openai.py
 devtorch_core/reasoning/__init__.py
 devtorch_core/reasoning/entry.py
@@ -217,6 +224,10 @@ tests/test_s25_delivery_time.py
 tests/test_s25_serve.py
 tests/test_s25_setup_detection.py
 tests/test_s25_templates.py
+tests/test_s26_daemon.py
+tests/test_s26_git_hooks.py
+tests/test_s26_proxy.py
+tests/test_s26_watcher.py
 tests/test_s2_context.py
 tests/test_s3_sensitivity.py
 tests/test_s4_invariants.py

{agentorch-2.2.0 → agentorch-2.3.0}/devtorch_cli/main.py

@@ -827,6 +827,61 @@ def cmd_proxy_status(args: argparse.Namespace) -> int:
     return 0
 
 
+# S26: daemon subcommand
+
+
+def cmd_daemon_start(args: argparse.Namespace) -> int:
+    from devtorch_core.daemon import start_daemon
+
+    project_root = Path(args.path).resolve()
+    ok, message = start_daemon(
+        project_root,
+        run_server=getattr(args, "run_server", False),
+        host=getattr(args, "host", "127.0.0.1"),
+        port=int(getattr(args, "port", 8765)),
+    )
+    print(message)
+    return 0 if ok else 1
+
+
+def cmd_daemon_stop(args: argparse.Namespace) -> int:
+    from devtorch_core.daemon import stop_daemon
+
+    project_root = Path(args.path).resolve()
+    ok, message = stop_daemon(project_root)
+    print(message)
+    return 0 if ok else 1
+
+
+def cmd_daemon_status(args: argparse.Namespace) -> int:
+    from devtorch_core.daemon import daemon_status
+
+    project_root = Path(args.path).resolve()
+    status = daemon_status(project_root)
+    if status.get("error"):
+        print(f"error: {status['error']}", file=sys.stderr)
+        return 1
+    if status["running"]:
+        print(f"DevTorch daemon running (PID {status['pid']})")
+    else:
+        print("DevTorch daemon stopped")
+    print(f"Log: {status['log_path']}")
+    return 0
+
+
+def cmd_daemon_logs(args: argparse.Namespace) -> int:
+    from devtorch_core.daemon import daemon_logs
+
+    project_root = Path(args.path).resolve()
+    lines = daemon_logs(project_root, lines=getattr(args, "lines", 50))
+    if not lines:
+        print("(no daemon log entries)")
+        return 0
+    for line in lines:
+        print(line)
+    return 0
+
+
 # Sprint 9: event log signing
 
 def cmd_signing_enable(args: argparse.Namespace) -> int:
@@ -1899,6 +1954,37 @@ def build_parser() -> argparse.ArgumentParser:
     p_proxy_status = proxy_sub.add_parser("status", help="Check if the proxy is running on :8765.")
     p_proxy_status.set_defaults(func=cmd_proxy_status)
 
+    # S26: daemon subcommand
+    p_daemon = subparsers.add_parser(
+        "daemon", help="Background file watcher and server (S26)."
+    )
+    daemon_sub = p_daemon.add_subparsers(dest="daemon_command", required=True)
+
+    p_daemon_start = daemon_sub.add_parser("start", help="Start the background daemon.")
+    p_daemon_start.add_argument(
+        "--run-server", action="store_true", dest="run_server",
+        help="Also start the unified DevTorch server in the daemon process.",
+    )
+    p_daemon_start.add_argument(
+        "--host", default="127.0.0.1", help="Host to bind the unified server to (default: 127.0.0.1)."
+    )
+    p_daemon_start.add_argument(
+        "--port", type=int, default=8765, help="Port to bind the unified server to (default: 8765)."
+    )
+    p_daemon_start.set_defaults(func=cmd_daemon_start)
+
+    p_daemon_stop = daemon_sub.add_parser("stop", help="Stop the background daemon.")
+    p_daemon_stop.set_defaults(func=cmd_daemon_stop)
+
+    p_daemon_status = daemon_sub.add_parser("status", help="Show daemon status.")
+    p_daemon_status.set_defaults(func=cmd_daemon_status)
+
+    p_daemon_logs = daemon_sub.add_parser("logs", help="Print the daemon log.")
+    p_daemon_logs.add_argument(
+        "--lines", type=int, default=50, help="Number of lines to show (default: 50)."
+    )
+    p_daemon_logs.set_defaults(func=cmd_daemon_logs)
+
     # S17a: calibrate subcommand
     p_calibrate = subparsers.add_parser("calibrate", help="Calibrate DevTorch metrics baselines.")
     p_calibrate.add_argument(
@@ -2081,6 +2167,22 @@ def build_parser() -> argparse.ArgumentParser:
     p_hooks_git_commit.add_argument("commit_msg_file", help="Path to the git commit message file.")
     p_hooks_git_commit.set_defaults(hooks_command="git-commit")
 
+    p_hooks_git_capture = hooks_sub.add_parser(
+        "git-capture",
+        help="Git capture hook entry point for post-commit, post-merge, post-checkout, post-rewrite.",
+    )
+    p_hooks_git_capture.add_argument(
+        "event",
+        choices=["post-commit", "post-merge", "post-checkout", "post-rewrite", "conflict"],
+        help="Git hook event to capture.",
+    )
+    p_hooks_git_capture.add_argument(
+        "args",
+        nargs="*",
+        help="Extra positional arguments passed by the git hook (e.g. squash flag, prev/new HEAD).",
+    )
+    p_hooks_git_capture.set_defaults(hooks_command="git-capture")
+
     p_hooks_run = hooks_sub.add_parser("run", help="Run a Claude Code hook (reads JSON from stdin).")
     p_hooks_run.add_argument("hook_name", choices=["pre-tool-use", "post-tool-use", "session-end"])
     p_hooks_run.set_defaults(hooks_command="run")
@@ -2520,6 +2622,10 @@ def main(argv: list[str] | None = None) -> int:
             result = install_all_hooks(project_root)
             print(f"Claude hooks: {'✓' if result.claude_hooks_written else '✗'} {result.claude_hooks_path}")
             print(f"Git hook:     {'✓' if result.git_hook_written else '✗'} {result.git_hook_path}")
+            if getattr(result, "git_capture_hooks_written", False):
+                print(f"Git capture hooks: ✓ ({len(result.git_capture_hooks)} installed)")
+            else:
+                print(f"Git capture hooks: ✗")
             if result.errors:
                 for e in result.errors:
                     print(f"  Error: {e}")
@@ -2530,6 +2636,9 @@ def main(argv: list[str] | None = None) -> int:
         elif args.hooks_command == "git-commit":
             from devtorch_core.hooks.git_commit import handle_commit_msg
             handle_commit_msg(args.commit_msg_file)
+        elif args.hooks_command == "git-capture":
+            from devtorch_core.hooks.git_capture import handle_git_capture
+            handle_git_capture(args.event, args.args)
         elif args.hooks_command == "run":
             from devtorch_core.hooks.runner import run as run_hook
             run_hook(args.hook_name)

agentorch-2.3.0/devtorch_core/daemon/__init__.py

@@ -0,0 +1,27 @@
+"""DevTorch background daemon (S26).
+
+Provides a long-running filesystem watcher that emits `fs_change` events to
+`.GCC/events.log.jsonl` without requiring IDE hooks.
+"""
+
+from devtorch_core.daemon.supervisor import (
+    PID_FILE_NAME,
+    LOG_FILE_NAME,
+    start_daemon,
+    stop_daemon,
+    daemon_status,
+    daemon_logs,
+    _run_daemon,
+)
+from devtorch_core.daemon.watcher import ProjectWatcher
+
+__all__ = [
+    "PID_FILE_NAME",
+    "LOG_FILE_NAME",
+    "start_daemon",
+    "stop_daemon",
+    "daemon_status",
+    "daemon_logs",
+    "_run_daemon",
+    "ProjectWatcher",
+]