From 317169cb310e9d088fbc01254ed83216fb763e3c Mon Sep 17 00:00:00 2001 From: thePR0M3TH3AN <53631862+PR0M3TH3AN@users.noreply.github.com> Date: Sat, 14 Feb 2026 17:59:10 -0500 Subject: [PATCH] Align scheduler prompts to scheduler-owned completion model --- src/prompts/META_PROMPTS.md | 8 ++++++-- src/prompts/daily/audit-agent.md | 2 +- src/prompts/daily/ci-health-agent.md | 2 +- src/prompts/daily/const-refactor-agent.md | 2 +- src/prompts/daily/content-audit-agent.md | 2 +- src/prompts/daily/decompose-agent.md | 2 +- src/prompts/daily/deps-security-agent.md | 2 +- src/prompts/daily/design-system-audit-agent.md | 2 +- src/prompts/daily/docs-agent.md | 2 +- src/prompts/daily/docs-alignment-agent.md | 2 +- src/prompts/daily/docs-code-investigator.md | 2 +- src/prompts/daily/innerhtml-migration-agent.md | 2 +- src/prompts/daily/known-issues-agent.md | 2 +- src/prompts/daily/load-test-agent.md | 2 +- src/prompts/daily/log-fixer-agent.md | 10 ++++------ src/prompts/daily/onboarding-audit-agent.md | 2 +- src/prompts/daily/perf-agent.md | 2 +- src/prompts/daily/protocol-research-agent.md | 2 +- src/prompts/daily/scheduler-update-agent.md | 2 +- src/prompts/daily/style-agent.md | 2 +- src/prompts/daily/test-audit-agent.md | 2 +- src/prompts/daily/todo-triage-agent.md | 2 +- .../daily/torch-garbage-collection-agent.md | 2 +- src/prompts/scheduler-flow.md | 14 +++++++------- src/prompts/weekly/bug-reproducer-agent.md | 2 +- src/prompts/weekly/changelog-agent.md | 2 +- src/prompts/weekly/dead-code-agent.md | 2 +- src/prompts/weekly/feature-proposer-agent.md | 2 +- src/prompts/weekly/frontend-console-debug-agent.md | 2 +- src/prompts/weekly/fuzz-agent.md | 2 +- src/prompts/weekly/perf-deepdive-agent.md | 2 +- src/prompts/weekly/perf-optimization-agent.md | 2 +- src/prompts/weekly/pr-review-agent.md | 2 +- src/prompts/weekly/prompt-maintenance-agent.md | 2 +- src/prompts/weekly/prompt-safety-agent.md | 2 +- src/prompts/weekly/race-condition-agent.md | 2 +- src/prompts/weekly/refactor-agent.md | 2 +- src/prompts/weekly/repo-fit-agent.md | 2 +- src/prompts/weekly/smoke-agent.md | 2 +- src/prompts/weekly/telemetry-agent.md | 2 +- src/prompts/weekly/test-coverage-agent.md | 2 +- src/prompts/weekly/ui-ux-agent.md | 2 +- src/prompts/weekly/weekly-synthesis-agent.md | 2 +- 43 files changed, 57 insertions(+), 55 deletions(-) diff --git a/src/prompts/META_PROMPTS.md b/src/prompts/META_PROMPTS.md index 627a8fb..61e219d 100644 --- a/src/prompts/META_PROMPTS.md +++ b/src/prompts/META_PROMPTS.md @@ -9,6 +9,8 @@ Copy one block below into the scheduler agent session. ```text You are the daily agent scheduler for this repository. +Authority model: **scheduler-owned completion/logging**. Spawned agents must not run `lock:complete` and must not write final `*_completed.md` / `*_failed.md` task logs. + Follow `src/prompts/scheduler-flow.md` exactly. MUST 1: Set cadence config to: @@ -49,7 +51,7 @@ MUST 3: Run these commands in this order: 5) Claim via repository lock: AGENT_PLATFORM= npm run lock:lock -- --agent --cadence daily Exit 0 = lock acquired, proceed. Exit 3 = race lost, go back to step 3. -6) Execute selected prompt from src/prompts/daily/ +6) Execute selected prompt from src/prompts/daily/ (spawned agent work only; completion publish and final task log writing remain scheduler-owned) 7) Run required memory workflow for this cadence: - Before execution, run the retrieval command if configured: `scheduler.memoryPolicyByCadence.daily.retrieveCommand` @@ -91,6 +93,8 @@ MUST 4: If all daily agents are excluded, stop and write `_failed.md` with this ```text You are the weekly agent scheduler for this repository. +Authority model: **scheduler-owned completion/logging**. Spawned agents must not run `lock:complete` and must not write final `*_completed.md` / `*_failed.md` task logs. + Follow `src/prompts/scheduler-flow.md` exactly. MUST 1: Set cadence config to: @@ -131,7 +135,7 @@ MUST 3: Run these commands in this order: 5) Claim via repository lock: AGENT_PLATFORM= npm run lock:lock -- --agent --cadence weekly Exit 0 = lock acquired, proceed. Exit 3 = race lost, go back to step 3. -6) Execute selected prompt from src/prompts/weekly/ +6) Execute selected prompt from src/prompts/weekly/ (spawned agent work only; completion publish and final task log writing remain scheduler-owned) 7) Run required memory workflow for this cadence: - Before execution, run the retrieval command if configured: `scheduler.memoryPolicyByCadence.weekly.retrieveCommand` diff --git a/src/prompts/daily/audit-agent.md b/src/prompts/daily/audit-agent.md index aa00998..1c24ce7 100644 --- a/src/prompts/daily/audit-agent.md +++ b/src/prompts/daily/audit-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **audit-agent**, a senior software engineer AI agent working inside this repositorysitory (target: default working branch). diff --git a/src/prompts/daily/ci-health-agent.md b/src/prompts/daily/ci-health-agent.md index e87e5b0..1acd493 100644 --- a/src/prompts/daily/ci-health-agent.md +++ b/src/prompts/daily/ci-health-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **ci-health-agent**, a senior software engineer agent working inside this repository. diff --git a/src/prompts/daily/const-refactor-agent.md b/src/prompts/daily/const-refactor-agent.md index c81e986..da5f1cb 100644 --- a/src/prompts/daily/const-refactor-agent.md +++ b/src/prompts/daily/const-refactor-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **const-refactor-agent**, a senior software engineer AI agent working inside this repository (target branch: default branch). diff --git a/src/prompts/daily/content-audit-agent.md b/src/prompts/daily/content-audit-agent.md index 18374f0..1fcd82f 100644 --- a/src/prompts/daily/content-audit-agent.md +++ b/src/prompts/daily/content-audit-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **content-audit-agent**, a senior user-docs verification AI engineer working inside this repositorysitory (target branch: default branch). diff --git a/src/prompts/daily/decompose-agent.md b/src/prompts/daily/decompose-agent.md index 28d8000..ad8f4ca 100644 --- a/src/prompts/daily/decompose-agent.md +++ b/src/prompts/daily/decompose-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **decompose-agent**, a senior software engineer AI agent working inside this repositorysitory (target branch: default branch). diff --git a/src/prompts/daily/deps-security-agent.md b/src/prompts/daily/deps-security-agent.md index 9e64bd9..c1a3ba7 100644 --- a/src/prompts/daily/deps-security-agent.md +++ b/src/prompts/daily/deps-security-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **deps-security-agent**, a senior software engineer AI agent operating inside the this repository (default branch). diff --git a/src/prompts/daily/design-system-audit-agent.md b/src/prompts/daily/design-system-audit-agent.md index 34cb5f3..924ef9e 100644 --- a/src/prompts/daily/design-system-audit-agent.md +++ b/src/prompts/daily/design-system-audit-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **design-system-audit-agent**, a senior front-end engineer agent working inside this repository. diff --git a/src/prompts/daily/docs-agent.md b/src/prompts/daily/docs-agent.md index 8442a39..e51684c 100644 --- a/src/prompts/daily/docs-agent.md +++ b/src/prompts/daily/docs-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **docs-agent**, a senior software engineer + documentation maintainer working inside this repository. diff --git a/src/prompts/daily/docs-alignment-agent.md b/src/prompts/daily/docs-alignment-agent.md index 16d686e..5ae8aad 100644 --- a/src/prompts/daily/docs-alignment-agent.md +++ b/src/prompts/daily/docs-alignment-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **docs-alignment-agent**, a senior documentation-alignment engineer working inside this repository. diff --git a/src/prompts/daily/docs-code-investigator.md b/src/prompts/daily/docs-code-investigator.md index 824471e..e410d59 100644 --- a/src/prompts/daily/docs-code-investigator.md +++ b/src/prompts/daily/docs-code-investigator.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **docs-code-investigator**, a senior software engineer AI agent working inside this repository (target branch: default branch). diff --git a/src/prompts/daily/innerhtml-migration-agent.md b/src/prompts/daily/innerhtml-migration-agent.md index 0e8ee10..a8e021c 100644 --- a/src/prompts/daily/innerhtml-migration-agent.md +++ b/src/prompts/daily/innerhtml-migration-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **innerhtml-migration-agent**, a senior software engineer AI agent working inside this repository (target branch: default branch). diff --git a/src/prompts/daily/known-issues-agent.md b/src/prompts/daily/known-issues-agent.md index 73e5597..3911a84 100644 --- a/src/prompts/daily/known-issues-agent.md +++ b/src/prompts/daily/known-issues-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **known-issues-agent**, a senior software engineer agent working inside this repository. diff --git a/src/prompts/daily/load-test-agent.md b/src/prompts/daily/load-test-agent.md index d95325f..0f23072 100644 --- a/src/prompts/daily/load-test-agent.md +++ b/src/prompts/daily/load-test-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **load-test-agent**, a senior performance engineer agent working inside this repository. diff --git a/src/prompts/daily/log-fixer-agent.md b/src/prompts/daily/log-fixer-agent.md index e37725b..461509a 100644 --- a/src/prompts/daily/log-fixer-agent.md +++ b/src/prompts/daily/log-fixer-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **log-fixer-agent**, a senior software engineer AI agent working inside this repository. @@ -20,7 +20,7 @@ PRIMARY GOALS / SUCCESS CRITERIA - Analyze the failure reason (e.g., syntax error, missing dependency, test failure, lock error). - Attempt to fix the issue if it is a code or configuration error. - Create a PR with the fix or an issue if the fix is complex/risky. -- Log your actions in `task-logs/daily/__log-fixer-agent__completed.md`. +- Provide a concise handoff summary in your response so the scheduler can write the final task log. =============================================================================== HARD CONSTRAINTS @@ -63,13 +63,11 @@ DAILY WORKFLOW - **Issue:** If the fix is too complex, risky, or unclear, create an issue. - Title: `issue(log-fixer): investigate failure` - Body: Details from the log, analysis, and potential solutions. - - **Log:** Create a log file `task-logs/daily/__log-fixer-agent__completed.md`. - - Summarize the failures found. - - List the actions taken (PRs created, issues opened, fixes applied). + - **Handoff Summary:** Include a final summary for the scheduler containing failures found and actions taken (PRs created, issues opened, fixes applied). =============================================================================== OUTPUTS -- `task-logs/daily/__log-fixer-agent__completed.md` +- Scheduler-managed final task log (written by scheduler after validation/completion publish). - PRs for code fixes. - Issues for complex problems. diff --git a/src/prompts/daily/onboarding-audit-agent.md b/src/prompts/daily/onboarding-audit-agent.md index 40d983f..6afeacc 100644 --- a/src/prompts/daily/onboarding-audit-agent.md +++ b/src/prompts/daily/onboarding-audit-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **onboarding-audit-agent**, a senior developer-experience (DX) engineer working inside this repository. diff --git a/src/prompts/daily/perf-agent.md b/src/prompts/daily/perf-agent.md index b90d2d8..849603c 100644 --- a/src/prompts/daily/perf-agent.md +++ b/src/prompts/daily/perf-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **perf-agent**, a senior software engineer AI agent working inside this repository (default branch). diff --git a/src/prompts/daily/protocol-research-agent.md b/src/prompts/daily/protocol-research-agent.md index c70dd77..11b7b3c 100644 --- a/src/prompts/daily/protocol-research-agent.md +++ b/src/prompts/daily/protocol-research-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **protocol-research-agent**, a senior engineer focused on standards and interoperability in this repository. diff --git a/src/prompts/daily/scheduler-update-agent.md b/src/prompts/daily/scheduler-update-agent.md index bcd0f65..e0caa8c 100644 --- a/src/prompts/daily/scheduler-update-agent.md +++ b/src/prompts/daily/scheduler-update-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **scheduler-update-agent**, a housekeeping agent working inside this repository. diff --git a/src/prompts/daily/style-agent.md b/src/prompts/daily/style-agent.md index d58cceb..b5f0afd 100644 --- a/src/prompts/daily/style-agent.md +++ b/src/prompts/daily/style-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **style-agent**, a senior software engineer agent working inside this repository. diff --git a/src/prompts/daily/test-audit-agent.md b/src/prompts/daily/test-audit-agent.md index 56b1a7d..339dd2b 100644 --- a/src/prompts/daily/test-audit-agent.md +++ b/src/prompts/daily/test-audit-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **test-audit-agent**, a senior software engineer AI agent working inside this repository (default branch). diff --git a/src/prompts/daily/todo-triage-agent.md b/src/prompts/daily/todo-triage-agent.md index 570add0..4794dae 100644 --- a/src/prompts/daily/todo-triage-agent.md +++ b/src/prompts/daily/todo-triage-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **todo-triage-agent**, a senior software engineer agent working inside this repository. diff --git a/src/prompts/daily/torch-garbage-collection-agent.md b/src/prompts/daily/torch-garbage-collection-agent.md index c1537c2..0c039f7 100644 --- a/src/prompts/daily/torch-garbage-collection-agent.md +++ b/src/prompts/daily/torch-garbage-collection-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **torch-garbage-collection-agent**, a repository hygiene agent focused on removing stale log files from TORCH. diff --git a/src/prompts/scheduler-flow.md b/src/prompts/scheduler-flow.md index 83e67c2..24d8a97 100644 --- a/src/prompts/scheduler-flow.md +++ b/src/prompts/scheduler-flow.md @@ -42,10 +42,10 @@ Every agent prompt invoked by the schedulers (daily/weekly) MUST enforce this co - Storage command MUST write cadence-scoped evidence artifacts: - `.scheduler-memory/store-.ok` - `.scheduler-memory/store-.json` containing operation inputs/outputs (`agentId`, input event count, stored count, generated summaries). -6. **Publish lock completion only after validation passes and before writing success logs**: - - Run `npm run lock:complete -- --agent --cadence ` (or equivalent `complete`) successfully - - If any validation command exits non-zero, do **not** call `lock:complete`; write `_failed.md` with the validation failure reason and stop - - Only after completion publish succeeds may the agent write final `*_completed.md` log files +6. **Scheduler-owned completion/logging is mandatory**: + - Spawned agents MUST NOT run `lock:complete`. + - Spawned agents MUST NOT write final `task-logs//____completed.md` or `__failed.md` files. + - Scheduler performs completion publish and writes final success/failure task logs after its own validation gates. ## Numbered MUST Procedure @@ -173,7 +173,7 @@ Every agent prompt invoked by the schedulers (daily/weekly) MUST enforce this co - If any validation command exits non-zero: **fail the run immediately**, write `_failed.md` with the failing command and reason, and stop. - When step 11 fails, step 12 MUST NOT be executed (`lock:complete` is forbidden until validation passes). -12. Publish completion before writing final success log: +12. Publish completion before writing final success log (scheduler-owned): ```bash AGENT_PLATFORM= \ @@ -185,10 +185,10 @@ Every agent prompt invoked by the schedulers (daily/weekly) MUST enforce this co - Exit `0`: completion published successfully; continue to step 13. - Exit non-zero: **fail the run**, write `_failed.md` with a clear reason that completion publish failed and retry guidance (for example: `Retry npm run lock:complete -- --agent --cadence after verifying relay connectivity`), then stop. -13. Create final task log only after step 12 succeeds: +13. Create final task log only after step 12 succeeds (scheduler-owned): - `_completed.md` MUST be created only after completion publish succeeds. - - `_failed.md` is required when step 10 or step 11 fails, and should include the failure reason and next retry action. + - `_failed.md` is required when step 10, step 11, or step 12 fails, and should include the failure reason and next retry action. 14. Commit/push behavior is delegated outside this scheduler script. diff --git a/src/prompts/weekly/bug-reproducer-agent.md b/src/prompts/weekly/bug-reproducer-agent.md index 778d4c0..9412238 100644 --- a/src/prompts/weekly/bug-reproducer-agent.md +++ b/src/prompts/weekly/bug-reproducer-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **bug-reproducer-agent**, a senior debugging engineer working inside this repository. diff --git a/src/prompts/weekly/changelog-agent.md b/src/prompts/weekly/changelog-agent.md index 7158597..d2914a9 100644 --- a/src/prompts/weekly/changelog-agent.md +++ b/src/prompts/weekly/changelog-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **changelog-agent**, a senior release-note editor and repo historian working inside this repository. diff --git a/src/prompts/weekly/dead-code-agent.md b/src/prompts/weekly/dead-code-agent.md index eac0607..48e995b 100644 --- a/src/prompts/weekly/dead-code-agent.md +++ b/src/prompts/weekly/dead-code-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **dead-code-agent**, a senior maintenance engineer working inside this repository. diff --git a/src/prompts/weekly/feature-proposer-agent.md b/src/prompts/weekly/feature-proposer-agent.md index be6a2f5..db0502f 100644 --- a/src/prompts/weekly/feature-proposer-agent.md +++ b/src/prompts/weekly/feature-proposer-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **feature-proposer-agent**, a weekly agent dedicated to incrementally adding value to the project by proposing and implementing new features. diff --git a/src/prompts/weekly/frontend-console-debug-agent.md b/src/prompts/weekly/frontend-console-debug-agent.md index 5c29bb9..b60bebc 100644 --- a/src/prompts/weekly/frontend-console-debug-agent.md +++ b/src/prompts/weekly/frontend-console-debug-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **frontend-console-debug-agent**, a senior frontend engineer working inside this repository. diff --git a/src/prompts/weekly/fuzz-agent.md b/src/prompts/weekly/fuzz-agent.md index 01f403b..10bec2c 100644 --- a/src/prompts/weekly/fuzz-agent.md +++ b/src/prompts/weekly/fuzz-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **fuzz-agent**, a senior robustness and security-minded engineer working inside this repository. diff --git a/src/prompts/weekly/perf-deepdive-agent.md b/src/prompts/weekly/perf-deepdive-agent.md index cc2c851..df98e7e 100644 --- a/src/prompts/weekly/perf-deepdive-agent.md +++ b/src/prompts/weekly/perf-deepdive-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **perf-deepdive-agent**, a senior performance engineer working inside this repository. diff --git a/src/prompts/weekly/perf-optimization-agent.md b/src/prompts/weekly/perf-optimization-agent.md index 9382ea8..48351dd 100644 --- a/src/prompts/weekly/perf-optimization-agent.md +++ b/src/prompts/weekly/perf-optimization-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **perf-optimization-agent**, a senior performance engineer working inside this repository. diff --git a/src/prompts/weekly/pr-review-agent.md b/src/prompts/weekly/pr-review-agent.md index f8fb95a..3476466 100644 --- a/src/prompts/weekly/pr-review-agent.md +++ b/src/prompts/weekly/pr-review-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **pr-review-agent**, a senior code reviewer and CI sherpa working inside this repository. diff --git a/src/prompts/weekly/prompt-maintenance-agent.md b/src/prompts/weekly/prompt-maintenance-agent.md index 966dd93..aebf69e 100644 --- a/src/prompts/weekly/prompt-maintenance-agent.md +++ b/src/prompts/weekly/prompt-maintenance-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **prompt-maintenance-agent**, a weekly maintenance agent responsible for the health, accuracy, and alignment of the agent prompt library. diff --git a/src/prompts/weekly/prompt-safety-agent.md b/src/prompts/weekly/prompt-safety-agent.md index a46e946..8100f70 100644 --- a/src/prompts/weekly/prompt-safety-agent.md +++ b/src/prompts/weekly/prompt-safety-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **prompt-safety-agent**, a senior software engineer agent responsible for auditing the safety and resilience of all other agent prompts in this repository. diff --git a/src/prompts/weekly/race-condition-agent.md b/src/prompts/weekly/race-condition-agent.md index bc20b10..b9a1c13 100644 --- a/src/prompts/weekly/race-condition-agent.md +++ b/src/prompts/weekly/race-condition-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **race-condition-agent**, a senior concurrency and reliability engineer working inside this repository. diff --git a/src/prompts/weekly/refactor-agent.md b/src/prompts/weekly/refactor-agent.md index 3330fa0..8e1be43 100644 --- a/src/prompts/weekly/refactor-agent.md +++ b/src/prompts/weekly/refactor-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **refactor-agent**, a senior software engineer working inside this repository. diff --git a/src/prompts/weekly/repo-fit-agent.md b/src/prompts/weekly/repo-fit-agent.md index 113ca76..0e1e780 100644 --- a/src/prompts/weekly/repo-fit-agent.md +++ b/src/prompts/weekly/repo-fit-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **repo-fit-agent**, a weekly maintenance agent that keeps TORCH aligned with the downstream repository it is used in. diff --git a/src/prompts/weekly/smoke-agent.md b/src/prompts/weekly/smoke-agent.md index 15c7c36..441fde2 100644 --- a/src/prompts/weekly/smoke-agent.md +++ b/src/prompts/weekly/smoke-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **smoke-agent**, a senior QA / integration engineer working inside this repository. diff --git a/src/prompts/weekly/telemetry-agent.md b/src/prompts/weekly/telemetry-agent.md index c1c3bc5..c4b7fe2 100644 --- a/src/prompts/weekly/telemetry-agent.md +++ b/src/prompts/weekly/telemetry-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **telemetry-agent**, a senior SRE/engineering-ops AI agent working inside this repository. diff --git a/src/prompts/weekly/test-coverage-agent.md b/src/prompts/weekly/test-coverage-agent.md index f49fc9d..fcfae0b 100644 --- a/src/prompts/weekly/test-coverage-agent.md +++ b/src/prompts/weekly/test-coverage-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **test-coverage-agent**, a senior test engineer & reliability maintainer working inside this repository. diff --git a/src/prompts/weekly/ui-ux-agent.md b/src/prompts/weekly/ui-ux-agent.md index ae9c1ad..290ff45 100644 --- a/src/prompts/weekly/ui-ux-agent.md +++ b/src/prompts/weekly/ui-ux-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **ui-ux-agent**, a UI/UX and marketing expert agent working inside this repository. diff --git a/src/prompts/weekly/weekly-synthesis-agent.md b/src/prompts/weekly/weekly-synthesis-agent.md index 770bdcc..fa85a76 100644 --- a/src/prompts/weekly/weekly-synthesis-agent.md +++ b/src/prompts/weekly/weekly-synthesis-agent.md @@ -6,7 +6,7 @@ - Run artifacts (required): update or explicitly justify omission for `src/context/`, `src/todo/`, `src/decisions/`, and `src/test_logs/`. - Unresolved issue handling (required): if unresolved/reproducible findings remain, update `KNOWN_ISSUES.md` and add or update an incidents note in `docs/agent-handoffs/incidents/`. - Memory contract (required): execute configured memory retrieval before implementation and configured memory storage after implementation, preserving scheduler evidence markers/artifacts. -- Completion ordering (required): run `npm run lock:complete -- --agent --cadence ` successfully before writing final `*_completed.md`; if validation fails, write `*_failed.md` and do not call `lock:complete`. +- Completion ownership (required): **do not** run `lock:complete` and **do not** create final `task-logs//____completed.md` or `__failed.md`; spawned agents hand results back to the scheduler, and the scheduler owns completion publishing/logging. You are: **weekly-synthesis-agent**, a senior engineering program assistant working inside this repository.