code-mower 1.0.8__tar.gz → 1.0.10__tar.gz

{code_mower-1.0.8 → code_mower-1.0.10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: code-mower
-Version: 1.0.8
+Version: 1.0.10
 Summary: Multi-reviewer AI code audit orchestration
 License-Expression: Apache-2.0
 Project-URL: Homepage, https://github.com/codemower-ai/code-mower
@@ -40,8 +40,7 @@ gate.
 
 The short version:
 
-- create safe, manual-first reviewer lanes for Codex, Claude, Gitar, and other
-  AI review tools;
+- create safe, manual-first peer-review lanes for Codex and Claude;
 - run setup diagnostics before a lane can surprise you with spend, source
   exposure, or GitHub Actions churn;
 - turn issue text, external docs, and project doctrine into local work orders
@@ -73,6 +72,18 @@ Code Mower should feel like an engineering tool, not a demo harness:
 
 ## What It Looks Like
 
+Start with Claude Code and Codex, or choose your participants during setup with
+`code-mower init --interactive`. For example, `code-mower init --with
+claude,codex,devin` previews that selection. The generated configuration remembers
+your choices for later sessions.
+
+Ask the agent you are using to start a Code Mower session; that agent is the
+default orchestrator. `code-mower session start --repo OWNER/REPO --with
+claude,codex,devin --host codex` prepares the operating brief when invoked from
+Codex. The host agent drives the work using available execution paths.
+See [Participants And Sessions](docs/sessions.md) for setup, host identity,
+and product-specific capabilities.
+
 `code-mower doctor --adoption --repo OWNER/REPO` is the first useful command
 for a real repository. It checks your runtime, GitHub setup, provider CLIs,
 token posture, optional cloud setup, private-repo Actions cost traps, and
@@ -92,7 +103,7 @@ Example, shortened:
 ```text
 $ code-mower doctor --adoption --repo OWNER/REPO
 PASS  config.validate             config validates
-PASS  profile.select              selected profile: codex, claude_audit, gitar
+PASS  profile.select              selected profile: codex, claude_audit
 PASS  runtime.python              Python 3.12 satisfies Code Mower requirements
 PASS  runtime.github_auth         GitHub CLI auth probe succeeded
 PASS  runtime.local_cli codex     codex found
@@ -205,6 +216,12 @@ Cold adopters should start with the reviewer gate, then add builders, then
 measure builder experiments; provider experiments come after that base loop is
 observable.
 
+The initial reviewer and builder paths default to Claude + Codex. Other
+participants require an explicit setup selection. `code-mower next-steps --repo
+OWNER/REPO --pr NUMBER` shows setup, peer-audit requests, status, and local
+evidence. Add `--advanced` for calibration, migration, package rehearsal, and
+optional cloud guidance after the initial loop works.
+
 Install first from the [Install And Bootstrap](docs/install.md) matrix: pipx for
 laptops, uv tool installs for hosted agents or CI boxes, and an editable venv
 for Code Mower contributors. All paths require Python 3.12 or newer. For
@@ -224,7 +241,7 @@ agent report its actual host posture, exact installed version, doctor result,
 Board URL, and any owner-only click-list before it mutates the repo.
 
 The current package-index announcement entry point is the tagged
-[Try Code Mower In 10 Minutes](https://github.com/codemower-ai/code-mower/blob/v1.0.8/docs/try-in-10-minutes.md)
+[Try Code Mower In 10 Minutes](https://github.com/codemower-ai/code-mower/blob/v1.0.10/docs/try-in-10-minutes.md)
 guide. The v1.0 supervised-pilot release includes the native Board, the
 controller dry-run and policy contract, adoption and upgrade hardening from
 recent install rehearsals, package-index release checks, provider-diversity
@@ -259,6 +276,15 @@ artifact, campaigns survive checkout changes, TestPyPI qualification is
 closed, Cursor builder identity is distinct from Grok review, Antigravity uses
 explicit project isolation, and hosted Cursor and Devin profiles require
 verified dispatch and result transport.
+The v1.0.9 patch makes Board refresh recovery deterministic, turns package
+installation failures into bounded actionable diagnostics, and lets release
+campaigns require established providers while retaining slower or experimental
+providers as informational evidence.
+The v1.0.10 patch makes local builder lanes deliverable and recoverable: a
+provider-neutral delivery contract with bounded no-change and owner-action
+outcomes, process-group cleanup so a finished provider is not reported as a
+timeout, simpler first-run defaults, host-led participant sessions, and a
+published 20-sample provider calibration scorecard.
 Package-only users can start from the public package rather than a source
 checkout.
 
@@ -291,7 +317,7 @@ Provider-contract baseline for the next release train:
 
 For release verification,
 [First-User Install Rehearsal](docs/first-user-install-rehearsal.md) records
-the package-index procedure for `v1.0.8` / `code-mower==1.0.8`. The
+the package-index procedure for `v1.0.10` / `code-mower==1.0.10`. The
 GitHub release records the workflow and rehearsal evidence for the exact tag.
 After publication, [Release Qualification](docs/release-qualification.md)
 coordinates provider-specific install and operational checks without treating
@@ -354,10 +380,11 @@ optional work order, single-writer branch, reviewer lanes, and fix rounds. Code
 Mower's templates now support that loop end to end; humans still own
 credentials, branch protection, calibration, and owner decisions.
 
-The reference adoption shape is Claude Code as orchestrator; Claude Code,
-Codex, and Cursor as builder lanes; Claude Code and Codex as reviewer lanes;
-and Gitar or Antigravity as informational reviewer signal until local
-calibration says otherwise.
+The initial adoption shape uses Claude Code or Codex as orchestrator, builder,
+and peer reviewer. The owning builder remains the only writer, and the other
+provider supplies independent review. Other builders and reviewers are explicit
+later additions; they follow the same lane policy with provider-specific
+execution and authentication where needed.
 
 ## Why Not Just Run Codex Or Claude Yourself?
 
@@ -439,10 +466,10 @@ The first recommended lanes are local/manual:
 | --- | --- | --- |
 | Codex audit | structured local peer audit | merge-gating eligible after setup |
 | Claude audit | structured local peer audit | merge-gating eligible after setup |
-| Gitar | advisory third signal | informational until calibrated |
 
-Everything else starts manual or informational until your own calibration data
-proves it is useful: Antigravity/Gemini, Muse, Hermes, CodeRabbit CLI, Cursor
+Everything else is opt-in after the Claude + Codex loop works and stays manual
+or informational until your own calibration data proves it is useful: Gitar,
+Antigravity/Gemini, Muse, Hermes, CodeRabbit CLI, Cursor
 BugBot, Qodo, Greptile, Devin, local LLMs, and future ACP bridges.
 
 Gemini CLI and Antigravity are distinct lane ids even though both are Google
@@ -480,8 +507,8 @@ measurement work.
 
 ## Installation Status
 
-The current package-index release baseline is `v1.0.8`, with pinned package
-install spec `code-mower==1.0.8`. Release evidence is recorded on the GitHub
+The current package-index release baseline is `v1.0.10`, with pinned package
+install spec `code-mower==1.0.10`. Release evidence is recorded on the GitHub
 release and in the first-user install rehearsal. The public repository is
 [codemower-ai/code-mower](https://github.com/codemower-ai/code-mower), and
 GitHub releases remain the auditable source for tags, build artifacts, and
@@ -516,8 +543,9 @@ first so local work exercises the same package entrypoint users install.
   conservative and team-controlled while automated retention jobs are roadmap
   work.
 - Advanced/provider/operator commands remain available behind
-  `code-mower --help-all`. The default help path stays focused on `init`,
-  `doctor`, calibration, value reports, and optional cloud export/upload.
+  `code-mower --help-all`. The default help path stays focused on setup,
+  `next-steps`, lane status, and local evidence; calibration and cloud commands
+  remain available through `--help-all`.
 
 ## Docs Map
 
@@ -537,6 +565,7 @@ first so local work exercises the same package entrypoint users install.
 - [Claude Lane Standing Instructions](docs/lanes/claude.md)
 - [Cursor Hosted Lane Standing Instructions](docs/lanes/cursor.md)
 - [Provider Matrix](docs/provider-matrix.md)
+- [Provider Calibration Scorecard](docs/provider-calibration-scorecard.md)
 - [GitHub Setup](docs/github-setup.md)
 - [First Run Transcript](docs/first-run-transcript.md)
 - [First-User Demo Transcript](docs/first-user-demo-transcript.md)
@@ -556,6 +585,8 @@ first so local work exercises the same package entrypoint users install.
 - [v1.0.6 Release Notes](docs/v106-release-notes.md)
 - [v1.0.7 Release Notes](docs/v107-release-notes.md)
 - [v1.0.8 Release Notes](docs/v108-release-notes.md)
+- [v1.0.9 Release Notes](docs/v109-release-notes.md)
+- [v1.0.10 Release Notes](docs/v1010-release-notes.md)
 - [Post-v0.8 Effectiveness Assessment](docs/post-v08-effectiveness-assessment.md)
 - [v1.0.1 Effectiveness Assessment](docs/v101-effectiveness-assessment.md)
 - [Demo Calibration Example](examples/demo-calibration/README.md)

{code_mower-1.0.8 → code_mower-1.0.10}/README.md

@@ -10,8 +10,7 @@ gate.
 
 The short version:
 
-- create safe, manual-first reviewer lanes for Codex, Claude, Gitar, and other
-  AI review tools;
+- create safe, manual-first peer-review lanes for Codex and Claude;
 - run setup diagnostics before a lane can surprise you with spend, source
   exposure, or GitHub Actions churn;
 - turn issue text, external docs, and project doctrine into local work orders
@@ -43,6 +42,18 @@ Code Mower should feel like an engineering tool, not a demo harness:
 
 ## What It Looks Like
 
+Start with Claude Code and Codex, or choose your participants during setup with
+`code-mower init --interactive`. For example, `code-mower init --with
+claude,codex,devin` previews that selection. The generated configuration remembers
+your choices for later sessions.
+
+Ask the agent you are using to start a Code Mower session; that agent is the
+default orchestrator. `code-mower session start --repo OWNER/REPO --with
+claude,codex,devin --host codex` prepares the operating brief when invoked from
+Codex. The host agent drives the work using available execution paths.
+See [Participants And Sessions](docs/sessions.md) for setup, host identity,
+and product-specific capabilities.
+
 `code-mower doctor --adoption --repo OWNER/REPO` is the first useful command
 for a real repository. It checks your runtime, GitHub setup, provider CLIs,
 token posture, optional cloud setup, private-repo Actions cost traps, and
@@ -62,7 +73,7 @@ Example, shortened:
 ```text
 $ code-mower doctor --adoption --repo OWNER/REPO
 PASS  config.validate             config validates
-PASS  profile.select              selected profile: codex, claude_audit, gitar
+PASS  profile.select              selected profile: codex, claude_audit
 PASS  runtime.python              Python 3.12 satisfies Code Mower requirements
 PASS  runtime.github_auth         GitHub CLI auth probe succeeded
 PASS  runtime.local_cli codex     codex found
@@ -175,6 +186,12 @@ Cold adopters should start with the reviewer gate, then add builders, then
 measure builder experiments; provider experiments come after that base loop is
 observable.
 
+The initial reviewer and builder paths default to Claude + Codex. Other
+participants require an explicit setup selection. `code-mower next-steps --repo
+OWNER/REPO --pr NUMBER` shows setup, peer-audit requests, status, and local
+evidence. Add `--advanced` for calibration, migration, package rehearsal, and
+optional cloud guidance after the initial loop works.
+
 Install first from the [Install And Bootstrap](docs/install.md) matrix: pipx for
 laptops, uv tool installs for hosted agents or CI boxes, and an editable venv
 for Code Mower contributors. All paths require Python 3.12 or newer. For
@@ -194,7 +211,7 @@ agent report its actual host posture, exact installed version, doctor result,
 Board URL, and any owner-only click-list before it mutates the repo.
 
 The current package-index announcement entry point is the tagged
-[Try Code Mower In 10 Minutes](https://github.com/codemower-ai/code-mower/blob/v1.0.8/docs/try-in-10-minutes.md)
+[Try Code Mower In 10 Minutes](https://github.com/codemower-ai/code-mower/blob/v1.0.10/docs/try-in-10-minutes.md)
 guide. The v1.0 supervised-pilot release includes the native Board, the
 controller dry-run and policy contract, adoption and upgrade hardening from
 recent install rehearsals, package-index release checks, provider-diversity
@@ -229,6 +246,15 @@ artifact, campaigns survive checkout changes, TestPyPI qualification is
 closed, Cursor builder identity is distinct from Grok review, Antigravity uses
 explicit project isolation, and hosted Cursor and Devin profiles require
 verified dispatch and result transport.
+The v1.0.9 patch makes Board refresh recovery deterministic, turns package
+installation failures into bounded actionable diagnostics, and lets release
+campaigns require established providers while retaining slower or experimental
+providers as informational evidence.
+The v1.0.10 patch makes local builder lanes deliverable and recoverable: a
+provider-neutral delivery contract with bounded no-change and owner-action
+outcomes, process-group cleanup so a finished provider is not reported as a
+timeout, simpler first-run defaults, host-led participant sessions, and a
+published 20-sample provider calibration scorecard.
 Package-only users can start from the public package rather than a source
 checkout.
 
@@ -261,7 +287,7 @@ Provider-contract baseline for the next release train:
 
 For release verification,
 [First-User Install Rehearsal](docs/first-user-install-rehearsal.md) records
-the package-index procedure for `v1.0.8` / `code-mower==1.0.8`. The
+the package-index procedure for `v1.0.10` / `code-mower==1.0.10`. The
 GitHub release records the workflow and rehearsal evidence for the exact tag.
 After publication, [Release Qualification](docs/release-qualification.md)
 coordinates provider-specific install and operational checks without treating
@@ -324,10 +350,11 @@ optional work order, single-writer branch, reviewer lanes, and fix rounds. Code
 Mower's templates now support that loop end to end; humans still own
 credentials, branch protection, calibration, and owner decisions.
 
-The reference adoption shape is Claude Code as orchestrator; Claude Code,
-Codex, and Cursor as builder lanes; Claude Code and Codex as reviewer lanes;
-and Gitar or Antigravity as informational reviewer signal until local
-calibration says otherwise.
+The initial adoption shape uses Claude Code or Codex as orchestrator, builder,
+and peer reviewer. The owning builder remains the only writer, and the other
+provider supplies independent review. Other builders and reviewers are explicit
+later additions; they follow the same lane policy with provider-specific
+execution and authentication where needed.
 
 ## Why Not Just Run Codex Or Claude Yourself?
 
@@ -409,10 +436,10 @@ The first recommended lanes are local/manual:
 | --- | --- | --- |
 | Codex audit | structured local peer audit | merge-gating eligible after setup |
 | Claude audit | structured local peer audit | merge-gating eligible after setup |
-| Gitar | advisory third signal | informational until calibrated |
 
-Everything else starts manual or informational until your own calibration data
-proves it is useful: Antigravity/Gemini, Muse, Hermes, CodeRabbit CLI, Cursor
+Everything else is opt-in after the Claude + Codex loop works and stays manual
+or informational until your own calibration data proves it is useful: Gitar,
+Antigravity/Gemini, Muse, Hermes, CodeRabbit CLI, Cursor
 BugBot, Qodo, Greptile, Devin, local LLMs, and future ACP bridges.
 
 Gemini CLI and Antigravity are distinct lane ids even though both are Google
@@ -450,8 +477,8 @@ measurement work.
 
 ## Installation Status
 
-The current package-index release baseline is `v1.0.8`, with pinned package
-install spec `code-mower==1.0.8`. Release evidence is recorded on the GitHub
+The current package-index release baseline is `v1.0.10`, with pinned package
+install spec `code-mower==1.0.10`. Release evidence is recorded on the GitHub
 release and in the first-user install rehearsal. The public repository is
 [codemower-ai/code-mower](https://github.com/codemower-ai/code-mower), and
 GitHub releases remain the auditable source for tags, build artifacts, and
@@ -486,8 +513,9 @@ first so local work exercises the same package entrypoint users install.
   conservative and team-controlled while automated retention jobs are roadmap
   work.
 - Advanced/provider/operator commands remain available behind
-  `code-mower --help-all`. The default help path stays focused on `init`,
-  `doctor`, calibration, value reports, and optional cloud export/upload.
+  `code-mower --help-all`. The default help path stays focused on setup,
+  `next-steps`, lane status, and local evidence; calibration and cloud commands
+  remain available through `--help-all`.
 
 ## Docs Map
 
@@ -507,6 +535,7 @@ first so local work exercises the same package entrypoint users install.
 - [Claude Lane Standing Instructions](docs/lanes/claude.md)
 - [Cursor Hosted Lane Standing Instructions](docs/lanes/cursor.md)
 - [Provider Matrix](docs/provider-matrix.md)
+- [Provider Calibration Scorecard](docs/provider-calibration-scorecard.md)
 - [GitHub Setup](docs/github-setup.md)
 - [First Run Transcript](docs/first-run-transcript.md)
 - [First-User Demo Transcript](docs/first-user-demo-transcript.md)
@@ -526,6 +555,8 @@ first so local work exercises the same package entrypoint users install.
 - [v1.0.6 Release Notes](docs/v106-release-notes.md)
 - [v1.0.7 Release Notes](docs/v107-release-notes.md)
 - [v1.0.8 Release Notes](docs/v108-release-notes.md)
+- [v1.0.9 Release Notes](docs/v109-release-notes.md)
+- [v1.0.10 Release Notes](docs/v1010-release-notes.md)
 - [Post-v0.8 Effectiveness Assessment](docs/post-v08-effectiveness-assessment.md)
 - [v1.0.1 Effectiveness Assessment](docs/v101-effectiveness-assessment.md)
 - [Demo Calibration Example](examples/demo-calibration/README.md)

{code_mower-1.0.8 → code_mower-1.0.10}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "code-mower"
-version = "1.0.8"
+version = "1.0.10"
 description = "Multi-reviewer AI code audit orchestration"
 requires-python = ">=3.12"
 readme = "README.md"

{code_mower-1.0.8 → code_mower-1.0.10}/src/code_mower/__init__.py

@@ -1,3 +1,3 @@
 """Code Mower package."""
 
-__version__ = "1.0.8"
+__version__ = "1.0.10"

{code_mower-1.0.8 → code_mower-1.0.10}/src/code_mower/board.py

@@ -297,6 +297,8 @@ def _format_timestamp(value: datetime) -> str:
 
 STATUS_CACHE_RETRY_BASE_SECONDS = 5.0
 STATUS_CACHE_RETRY_MAX_SECONDS = 60.0
+STATUS_CACHE_REFRESH_TIMEOUT_SECONDS = 120.0
+STATUS_CACHE_TIMEOUT_SECONDS = STATUS_CACHE_REFRESH_TIMEOUT_SECONDS
 
 
 class StatusCache:
@@ -314,16 +316,20 @@ class StatusCache:
     consumers can identify *which* completed snapshot they are holding
     independently of whether that snapshot is still fresh.
 
-    A failed refresh (or a failed refresh thread start) opens a bounded retry
-    backoff window. Without it the cached snapshot stays stale while
-    ``refresh_in_progress`` returns to false the instant the failure is
-    recorded, so the very next request would start another expensive
-    GitHub/local recomputation -- once per fast poll (~750ms) for as long as the
-    failure persists. During the window ``get()`` still answers immediately from
-    cold/stale metadata but starts nothing and reports ``refresh_in_progress``
-    false, so the browser drops back to its normal interval. The window doubles
-    per consecutive failure from ``retry_base_seconds`` up to
-    ``retry_max_seconds`` and is cleared by the first success.
+    A failed refresh (including abnormal thread exits or non-``Exception``
+    terminations), a dead/abandoned refresh that exceeds the recovery timeout,
+    or a failed refresh thread start opens a bounded retry backoff window.
+    Without it the cached snapshot stays stale while ``refresh_in_progress``
+    returns to false the instant the failure is recorded, so the very next
+    request would start another expensive GitHub/local recomputation -- once
+    per fast poll (~750ms) for as long as the failure persists. During the
+    window ``get()`` still answers immediately from cold/stale metadata but
+    starts nothing and reports ``refresh_in_progress`` false, so the browser
+    drops back to its normal interval. The window doubles per consecutive
+    failure from ``retry_base_seconds`` up to ``retry_max_seconds`` and is
+    cleared by the first success. Ownership tokens ensure that late-exiting or
+    abandoned threads cannot overwrite newer generations or clear the
+    in-progress flag of a replacement refresh.
     """
 
     def __init__(
@@ -336,6 +342,8 @@ class StatusCache:
         start_thread: Any = None,
         retry_base_seconds: float = STATUS_CACHE_RETRY_BASE_SECONDS,
         retry_max_seconds: float = STATUS_CACHE_RETRY_MAX_SECONDS,
+        refresh_timeout_seconds: float = STATUS_CACHE_REFRESH_TIMEOUT_SECONDS,
+        timeout_seconds: float | None = None,
     ) -> None:
         self._compute = compute
         self._ttl_seconds = max(float(ttl_seconds), 0.0)
@@ -344,11 +352,16 @@ class StatusCache:
         self._start_thread = start_thread or self._default_start_thread
         self._retry_base_seconds = max(float(retry_base_seconds), 0.0)
         self._retry_max_seconds = max(float(retry_max_seconds), self._retry_base_seconds)
+        if timeout_seconds is not None:
+            refresh_timeout_seconds = timeout_seconds
+        self._refresh_timeout_seconds = max(float(refresh_timeout_seconds), 0.0)
         self._lock = Lock()
         self._snapshot: dict[str, Any] | None = None
         self._computed_at: datetime | None = None
         self._computed_monotonic: float | None = None
         self._refreshing = False
+        self._refresh_started_monotonic: float | None = None
+        self._refresh_owner = 0
         self._generation = 0
         self._last_error: str = ""
         self._last_error_at: datetime | None = None
@@ -403,6 +416,8 @@ class StatusCache:
         so nothing here can leak paths, output, or credentials.
         """
         self._refreshing = False
+        self._refresh_started_monotonic = None
+        self._refresh_owner += 1
         self._last_error = _cache_error_summary(exc)
         self._last_error_at = self._now()
         self._consecutive_failures += 1
@@ -418,6 +433,13 @@ class StatusCache:
                 else None
             )
             is_stale = snapshot is None or age is None or age >= self._ttl_seconds
+            if (
+                self._refreshing
+                and self._refresh_started_monotonic is not None
+                and self._refresh_timeout_seconds > 0.0
+                and max(elapsed - self._refresh_started_monotonic, 0.0) >= self._refresh_timeout_seconds
+            ):
+                self._record_failure_locked(TimeoutError("status refresh timed out"))
             retry_in = (
                 max(self._retry_after_monotonic - elapsed, 0.0)
                 if self._retry_after_monotonic is not None
@@ -425,8 +447,12 @@ class StatusCache:
             )
             in_backoff = retry_in > 0.0
             should_start = is_stale and not self._refreshing and not in_backoff
+            owner: int | None = None
             if should_start:
                 self._refreshing = True
+                self._refresh_owner += 1
+                owner = self._refresh_owner
+                self._refresh_started_monotonic = elapsed
                 self._retry_after_monotonic = None
             metadata_snapshot = {
                 "state": "cold" if snapshot is None else ("stale" if is_stale else "fresh"),
@@ -439,35 +465,54 @@ class StatusCache:
                 "last_error": self._last_error,
                 "last_error_at": _format_timestamp(self._last_error_at) if self._last_error_at else "",
             }
-        if should_start:
+        if should_start and owner is not None:
+            def refresh_worker() -> None:
+                self._refresh(owner)
+
             try:
-                self._start_thread(self._refresh)
-            except Exception as exc:  # noqa: BLE001 - a failed thread start must never crash the endpoint
+                self._start_thread(refresh_worker)
+            except (Exception, SystemExit) as exc:  # noqa: BLE001 - a failed thread start must never crash the endpoint
                 with self._lock:
-                    self._record_failure_locked(exc)
-                    metadata_snapshot["refresh_in_progress"] = False
-                    metadata_snapshot["retry_in_seconds"] = round(self._retry_delay_locked(), 3)
-                    metadata_snapshot["last_error"] = self._last_error
-                    metadata_snapshot["last_error_at"] = _format_timestamp(self._last_error_at)
+                    if self._refresh_owner == owner:
+                        self._record_failure_locked(exc)
+                        metadata_snapshot["refresh_in_progress"] = False
+                        metadata_snapshot["retry_in_seconds"] = round(self._retry_delay_locked(), 3)
+                        metadata_snapshot["last_error"] = self._last_error
+                        metadata_snapshot["last_error_at"] = _format_timestamp(self._last_error_at)
         return snapshot, metadata_snapshot
 
-    def _refresh(self) -> None:
+    def _refresh(self, owner: int | None = None) -> None:
+        if owner is None:
+            with self._lock:
+                owner = self._refresh_owner
+        completed = False
         try:
             result = self._compute()
-        except Exception as exc:  # noqa: BLE001 - a background refresh must never crash the server
             with self._lock:
-                self._record_failure_locked(exc)
+                if self._refresh_owner == owner:
+                    self._snapshot = result
+                    self._computed_at = self._now()
+                    self._computed_monotonic = self._clock()
+                    self._generation += 1
+                    self._refreshing = False
+                    self._refresh_started_monotonic = None
+                    self._last_error = ""
+                    self._last_error_at = None
+                    self._consecutive_failures = 0
+                    self._retry_after_monotonic = None
+                    self._refresh_owner += 1
+                    completed = True
+        except BaseException as exc:  # noqa: BLE001 - a background refresh must never crash the server
+            with self._lock:
+                if self._refresh_owner == owner:
+                    self._record_failure_locked(exc)
+                    completed = True
             return
-        with self._lock:
-            self._snapshot = result
-            self._computed_at = self._now()
-            self._computed_monotonic = self._clock()
-            self._generation += 1
-            self._refreshing = False
-            self._last_error = ""
-            self._last_error_at = None
-            self._consecutive_failures = 0
-            self._retry_after_monotonic = None
+        finally:
+            if not completed:
+                with self._lock:
+                    if self._refresh_owner == owner and self._refreshing:
+                        self._record_failure_locked(RuntimeError("refresh thread exited abnormally"))
 
 
 def _pending_status_payload(config: BoardConfig) -> dict[str, Any]:
@@ -1370,7 +1415,7 @@ def render_board_html(config: BoardConfig) -> str:
       const campaigns = campaignsData.campaigns || [];
       const campaignRows = campaigns.flatMap(c => {{
         const header = `<div class="row"><div class="line"><b>Release ${{esc(c.release_tag)}}</b>${{pill(c.status)}}${{c.dry_run ? pill("dry-run") : pill("applied")}}${{pill(c.qualification_context)}}<span>${{seconds(c.elapsed_seconds)}}</span></div><div>next: <b>${{esc(c.next_action)}}</b></div></div>`;
-        const cardRows = (c.cards || []).map(card => `<div class="row" style="margin-left:16px"><div class="line"><b>${{esc(card.provider)}}</b><span class="${{stateClass(card.state)}}">${{esc(card.state)}}</span>${{pill(card.environment)}}${{card.transport_verified === false ? pill("transport unverified") : ""}}<span>${{seconds(card.elapsed_seconds)}}</span></div><div>next: <b>${{esc(card.next_action)}}</b></div>${{card.next_detail ? `<div class="muted">${{esc(card.next_detail)}}</div>` : ""}}${{card.response_deadline_at ? `<div class="muted" title="${{esc(card.response_deadline_at)}}">response deadline ${{localTime(card.response_deadline_at)}}</div>` : ""}}</div>`);
+        const cardRows = (c.cards || []).map(card => `<div class="row" style="margin-left:16px"><div class="line"><b>${{esc(card.provider)}}</b>${{pill(card.posture || "required")}}<span class="${{stateClass(card.state)}}">${{esc(card.state)}}</span>${{pill(card.environment)}}${{card.transport_verified === false ? pill("transport unverified") : ""}}<span>${{seconds(card.elapsed_seconds)}}</span></div><div>next: <b>${{esc(card.next_action)}}</b></div>${{card.next_detail ? `<div class="muted">${{esc(card.next_detail)}}</div>` : ""}}${{card.response_deadline_at ? `<div class="muted" title="${{esc(card.response_deadline_at)}}">response deadline ${{localTime(card.response_deadline_at)}}</div>` : ""}}</div>`);
         return [header, ...cardRows];
       }}).join("");
       put("campaigns", campaignRows || empty(campaignsData.message || "No release campaigns."));
@@ -1815,14 +1860,16 @@ def make_handler(
     *,
     gh_json_runner: lane_status.GitHubJsonRunner = lane_status.run_gh_json,
     command_runner: lane_status.CommandRunner = lane_status.run_command,
+    status_cache: StatusCache | None = None,
 ) -> type[BaseHTTPRequestHandler]:
     last_recorded_at: datetime | None = None
     last_recorded_generation = 0
     recording_lock = Lock()
-    status_cache = StatusCache(
-        lambda: status_payload(config, gh_json_runner=gh_json_runner, command_runner=command_runner),
-        ttl_seconds=config.refresh_seconds,
-    )
+    if status_cache is None:
+        status_cache = StatusCache(
+            lambda: status_payload(config, gh_json_runner=gh_json_runner, command_runner=command_runner),
+            ttl_seconds=config.refresh_seconds,
+        )
 
     class BoardHandler(BaseHTTPRequestHandler):
         server_version = "CodeMowerBoard/0.1"

{code_mower-1.0.8 → code_mower-1.0.10}/src/code_mower/builder_runs.py

@@ -39,6 +39,8 @@ AUTHOR_INFERENCE_RULES = {
     "claude-bot": ("claude", "claude_code_action"),
     "cursor[bot]": ("cursor_cloud_agent", "cursor_cloud_agent"),
     "cursor": ("cursor_cloud_agent", "cursor_cloud_agent"),
+    "devin-ai-integration": ("devin", "devin"),
+    "devin-ai-integration[bot]": ("devin", "devin"),
 }
 BRANCH_PREFIX_INFERENCE_RULES = (
     ("cursor/", "cursor_cloud_agent", "cursor_cloud_agent"),
@@ -49,6 +51,8 @@ BRANCH_PREFIX_INFERENCE_RULES = (
     ("chatgpt-codex-", "codex", "chatgpt-codex-connector"),
     ("claude/", "claude", "claude_code_action"),
     ("claude-", "claude", "claude_code_action"),
+    ("devin/", "devin_cli", "devin_cli"),
+    ("devin-", "devin_cli", "devin_cli"),
 )