teddy-cli 0.1.17.dev1267__tar.gz → 0.1.17.dev1269__tar.gz

{teddy_cli-0.1.17.dev1267 → teddy_cli-0.1.17.dev1269}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: teddy-cli
-Version: 0.1.17.dev1267
+Version: 0.1.17.dev1269
 Summary: A local-first, file-based AI coding workflow that applies the UNIX philosophy to AI collaboration.
 Author: Raphael Atteritano
 License: AGPL-3.0-only

{teddy_cli-0.1.17.dev1267 → teddy_cli-0.1.17.dev1269}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "teddy-cli"
-version = "0.1.17.dev1267"
+version = "0.1.17.dev1269"
 description = "A local-first, file-based AI coding workflow that applies the UNIX philosophy to AI collaboration."
 authors = [{ name = "Raphael Atteritano" }]
 license = { text = "AGPL-3.0-only" }

{teddy_cli-0.1.17.dev1267 → teddy_cli-0.1.17.dev1269}/src/teddy_executor/resources/MRP.xml

@@ -52,7 +52,7 @@ NOTES:
       2. Commit Message: Use Conventional Commits '<type>(<scope>): <description>' wrapped in single quotes. Use imperative mood. Append ! for breaking changes.
       3. Execution: Run `make commit '<type>(<scope>): <description>'` to stage, run the local pre-commit hooks, and commit. `make commit` stages ALL local changes, including currently unstaged ones (`git add .`). The target commits WITHOUT `--no-verify`, so Git's own pre-commit hook is a real local gate; the post-commit test gate always runs and is unaffected.
       4. Post-conditions: All tests MUST pass — the post-commit hook runs the full suite and reverts the commit on failure; this gate is unskippable and MUST NEVER be bypassed. The pre-commit stage is advisory (its failures do not block the local commit), but if it reported non-trivial quality-gate issues (e.g. file length), you MUST make a follow-up cleanup commit that resolves them before resuming feature work. Triage any resulting `[DEBT]` per the Technical Debt Triage policy — there is no `## Technical Debt` holding pen.
-      5. Push at Conclusion: In the Conclusion phase, run `make push` to rebase onto the remote, push, and wait for and report the CI results. `make push` compares the pushed SHA's failing-job set against the base run (`origin/main` before the push), per workflow: base green → new red = YOUR regression → fix it immediately before concluding (exit 1); base red → new red (same jobs) = pre-existing → report it and continue (exit 0); all green → PASSED. An unknown base is treated as yours (fail-safe). Tests MUST always be fixed; quality-gate failures MUST be fixed or explicitly planned in before you conclude.
+      5. Push at Conclusion: In the Conclusion phase, run `make push` to rebase onto the remote, push, and report the CI results. `make push` answers ONE question only: did this push turn a BLOCKING gate red? The pushed RUN's own conclusion is authoritative — it already folds in `continue-on-error`, so a non-blocking job (e.g. the CI `Quality Checks` job) can NEVER block the session. Only when the run-level conclusion is `failure` does attribution run: the base run (`origin/main` before the push) is classified as green, red, job-less, or absent, and a base that exists but is job-less is NOT green. A failure your push introduced (base green/absent/job-less -> new red) is YOUR regression (exit 1, fix before concluding); a failure already on the base (base red -> new red) is pre-existing (report and continue). Tests MUST always be fixed. A non-blocking quality result MUST be triaged per the Technical Debt Triage policy — fixed in-flight or explicitly scheduled into a slice or milestone — and MUST NOT be silently accepted.
     </rule>
 
     <rule n="7" name="Standardized Plan Types">

{teddy_cli-0.1.17.dev1267 → teddy_cli-0.1.17.dev1269}/src/teddy_executor/resources/templates/makefile.md

@@ -128,14 +128,16 @@ commit:
 
 ## Push (Deferred to Conclusion)
 
-`make push` runs `git pull --rebase`, captures the base and the pushed SHA, pushes, then discovers **every** CI run for the pushed SHA (`gh run list --commit <sha>`) and waits for and reports each workflow with its jobs. It blocks only on failures **your push introduced**:
+`make push` runs `git pull --rebase`, captures the base and the pushed SHA, pushes, then discovers **every** CI run for the pushed SHA (`gh run list --commit <sha>`) and waits for and reports each workflow with its jobs. It answers ONE question: **did this push turn a BLOCKING gate red?**
 
-- **base green → new red** = your regression → **exit 1** (fix before concluding).
+The pushed RUN's own conclusion is authoritative: it already folds in `continue-on-error`, so a **non-blocking** job (e.g. the CI `Quality Checks` job) can NEVER block the session — it is reported as a triage item instead. Only when the run-level conclusion is `failure` does attribution run, comparing the pushed run's failing-JOB set against the base run's:
+
+- **base green/absent/job-less → new red** = your regression → **exit 1** (fix before concluding).
 - **base red → new red (same jobs)** = pre-existing → **report and exit 0** (do not wait, do not duplicate the fix).
-- **all green** → PASSED.
-- **unknown base** → fail-safe: treat as yours (block).
+- **all green / non-blocking** → PASSED → **exit 0**.
+- **job-less base** → NOT green: a base run that exists but carries zero jobs (an invalid-workflow failure) is treated as an unknown base and fails safe (block).
 
-Attribution is deterministic: SHA → run → per-job conclusion (never "the latest run").
+Attribution is deterministic: SHA → run → run-level + per-job conclusion (never "the latest run").
 
 Three behaviors make the attribution trustworthy:
 
@@ -172,14 +174,36 @@ push:
 	fi; \
 	for WF in $$WFS; do \
 	  NEW=$$(gh run list --commit "$$SHA" --workflow "$$WF" --json databaseId --jq '.[0].databaseId'); \
-	  OLD=$$(gh run list --commit "$$BASE" --workflow "$$WF" --json databaseId --jq '.[0].databaseId'); \
 	  gh run watch "$$NEW" --exit-status >/dev/null 2>&1 || true; \
+	  NEWCONC=$$(gh run view "$$NEW" --json conclusion --jq '.conclusion'); \
 	  NEWFAIL=$$(gh run view "$$NEW" --json jobs --jq '[.jobs[] | select(.conclusion=="failure") | .name] | sort | join(",")'); \
-	  if [ -z "$$NEWFAIL" ]; then echo "$$WF: PASSED"; continue; fi; \
-	  if [ -z "$$OLD" ]; then echo "$$WF: FAILED ($$NEWFAIL) — unknown base, treating as YOUR regression"; STATUS=1; continue; fi; \
+	  NEWSTEPS=$$(gh run view "$$NEW" --json jobs --jq '[.jobs[] | select(.conclusion=="failure") | .steps[]? | select(.conclusion=="failure") | .name] | sort | join(", ")'); \
+	  if [ -n "$$NEWSTEPS" ]; then echo "$$WF: failing step(s): $$NEWSTEPS"; fi; \
+	  if [ "$$NEWCONC" != "failure" ]; then \
+	    if [ -n "$$NEWFAIL" ]; then \
+	      echo "$$WF: PASSED (run=$$NEWCONC) — non-blocking failure(s) [$$NEWFAIL] — triage, does not block"; \
+	    else \
+	      echo "$$WF: PASSED"; \
+	    fi; \
+	    continue; \
+	  fi; \
+	  OLD=$$(gh run list --commit "$$BASE" --workflow "$$WF" --json databaseId --jq '.[0].databaseId'); \
+	  if [ -z "$$OLD" ]; then \
+	    echo "$$WF: BLOCKING failure ($$NEWFAIL) — base has no run (unknown base) — YOUR regression"; \
+	    STATUS=1; continue; \
+	  fi; \
+	  OLDJOBS=$$(gh run view "$$OLD" --json jobs --jq '.jobs | length'); \
+	  if [ "$$OLDJOBS" = "0" ]; then \
+	    echo "$$WF: BLOCKING failure ($$NEWFAIL) — base run is job-less (invalid workflow, not green) — YOUR regression"; \
+	    STATUS=1; continue; \
+	  fi; \
 	  OLDFAIL=$$(gh run view "$$OLD" --json jobs --jq '[.jobs[] | select(.conclusion=="failure") | .name] | sort | join(",")'); \
-	  if [ "$$NEWFAIL" = "$$OLDFAIL" ]; then echo "$$WF: pre-existing failure ($$NEWFAIL) — not yours"; \
-	  else echo "$$WF: NEW failure ($$NEWFAIL) — YOUR regression"; STATUS=1; fi; \
+	  if [ "$$NEWFAIL" = "$$OLDFAIL" ]; then \
+	    echo "$$WF: pre-existing BLOCKING failure ($$NEWFAIL) — not yours"; \
+	  else \
+	    echo "$$WF: BLOCKING failure ($$NEWFAIL) — YOUR regression"; \
+	    STATUS=1; \
+	  fi; \
 	done; \
 	exit $$STATUS
 ```

{teddy_cli-0.1.17.dev1267 → teddy_cli-0.1.17.dev1269}/src/teddy_executor/resources/templates/pre-commit.md

@@ -31,6 +31,8 @@ repos:
       - id: trailing-whitespace
       - id: end-of-file-fixer
       - id: check-toml
+      - id: check-yaml
+      - id: check-json
       - id: check-added-large-files
         args: ['--maxkb=500']
       - id: check-case-conflict