Consolidate housekeeper runtime entrypoints#27
Open
nxs9bg24js-tech wants to merge 71 commits intomainfrom
Open
Consolidate housekeeper runtime entrypoints#27nxs9bg24js-tech wants to merge 71 commits intomainfrom
nxs9bg24js-tech wants to merge 71 commits intomainfrom
Conversation
✅ 新增内容： 1. Self-Evolving Agent 自进化智能体（2,720 字） 2. DyTopo 动态拓扑网络革命（4,121 字） 3. 研究索引 README（1,973 字） 📊 核心发现： - 🚨 80亿参数"绞杀"1200亿参数 - ✅ 突破 Scaling Law 铁律 - 🔄 静态群聊 → 自由交易集市 🎯 技术价值： - 动态拓扑网络架构 - 上下文污染解决方案 - 自进化机制设计 📚 总字数：8,814 字 🔗 来源：YouTube 播放列表深度分析
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- task_contract: add bidirectional HousekeeperTaskStatus mapping - worker_registry: import worker_status_rank from unified worker_contract replacing inline status rank dict - Add 11 new HousekeeperTaskStatus roundtrip tests Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…nd 2) - Offline demo fixture with success/failure/review flows and demo runner script - Fake Linux and Windows/Yingdao worker adapters for offline simulation - Heartbeat/lease/timeout simulation and 10-category failure taxonomy - Exhaustive illegal state transition test matrix for TaskStatus and RunStatus - Five gate scenario tests and retry/fallback/needs_review rule tests - Enhanced console: task filtering, worker metrics, run detail with logs/artifacts, and approval actions (approve/reject/retry/fallback) - Add pytest-asyncio to requirements.lock for CI async test support - Update CI workflow paths to include all new test and source files Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Acceptance test script: 30 consecutive runs with 5 fault types (timeout, crash, overreach, missing_artifacts, permission_denied) - Weighted deterministic fault selection with ~25% success baseline - Per-fault-type breakdown in console report - JSON machine-readable report output (--json-report) - Pytest wrapper (test_acceptance_harness.py) for CI integration - Updated offline-demo.md with acceptance harness docs, console demo, run detail walkthrough, and review/retry/fallback decision logic - Updated CI to include new test paths Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Linux housekeeper supervisor script and systemd-compatible runner - dispatch_to_linux.sh for remote task dispatch - WORKER_CONTRACT_SPEC.md documenting unified worker interface - CLAUDE.md project instructions for AI-assisted development - .env.linux.example template for Linux worker configuration - Update Makefile with new targets - Update README and linux-remote-worker docs Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
… checklist - 5 acceptance gates (G1-G5): contract models, state machine legality, gate verdict correctness, 30-run harness, API availability - 5 fail-fast conditions (F1-F5): lint/format, import integrity, state machine invariants, gate core rules, acceptance harness - PR merge checklist: code quality, tests, contract sync, docs, architecture - Demo-ready minimum standard (DR1-DR5): offline demos, console interaction, contract tests, CI green, docs complete Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…integration migration plan Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…cklist - OFFLINE_CAPABILITY_AUDIT.md: 11 REAL / 8 OFFLINE-ONLY / 7 FAKE classification - test_acceptance_gaps.py: 8 blind-spot test classes (concurrent dispatch, heartbeat timeout, cancel flow, orphan recovery, artifact validation, approval pre-check, gate after DriverResult, max-concurrent enforcement) - test_worker_contract_compliance.py: 5 contract classes for worker certification (registration, heartbeat, lifecycle, gate, error classification) - WORKER_ONBOARDING_CHECKLIST.md: checklist for certifying new workers - CI paths updated with both new test files 475 tests passing (52 new). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…tput to unified contracts Six bridge functions translate LinuxSupervisorConclusion/TaskSummaryRead/HeartbeatRead into GateOutcome/GateCheck/RunStatus/WorkerHeartbeat/WorkerRegistration without modifying any production service or unified contract model. 43 tests cover all 7 conclusions through the full chain (summary → gate checks → verdict). CI paths and acceptance criteria updated. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…uction path After run_once() returns a LinuxSupervisorTaskSummaryRead, the control plane now calls bridge functions to produce GateOutcome, GateCheck[], RunStatus, and GateVerdict. These are stored in result_payload["gate_evaluation"]. Task status decision logic (summary.success → COMPLETED/FAILED) is unchanged — gate evaluation is observational, proving real Linux paths are constrained by unified contracts. 6 integration tests pass with real subprocess execution. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
8 integration tests proving that real Linux execution through ControlPlaneService produces run_record in result_payload with unified RunStatus/RunRecord compatible data. Tests cover SUCCEEDED, FAILED (timeout, infra_error), NEEDS_REVIEW (unknown conclusion), field consistency with summary, result_data bridge fields, and error_message propagation. CI paths and production path changelog updated. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…te() LINUX_SUPERVISOR branch After gate evaluation, call supervisor_summary_to_run_record() to produce a BridgeRunRecord with unified RunStatus. Store as result_payload["run_record"] with datetime fields serialized to ISO strings. Task status decision logic unchanged. 8 integration tests pass. queued_at/leased_at approximation: started_at from summary.started_at (completed_at from summary.finished_at). No separate lease mechanism exists in LinuxSupervisorService. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
asdict() on frozen dataclass preserves RunStatus enum in the 'status' field. Add explicit .value serialization so result_payload["run_record"]["status"] is always a plain string, consistent with gate_evaluation.run_status. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
ACCEPTANCE_CRITERIA.md: Add G7 gate (9 items) for production path integration — gate evaluation, run lifecycle, field consistency. WORKER_GAP_ANALYSIS.md: Mark gate/run as wired, add wiring status table showing 5/7 bridge functions connected, 2 (heartbeat/registration) still unwired. Document queued_at/leased_at approximation limitations. No production code changed. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Update docs to reflect Codex's two-commit heartbeat wiring (86de697 + 6307136): both get_worker_heartbeat() and list_workers() now use the bridge function and return consistent unified WorkerStatus. - ACCEPTANCE_CRITERIA.md: add G8 gate (7 conditions, all implemented) - WORKER_GAP_ANALYSIS.md: add Section 6 heartbeat audit, mark WorkerHeartbeat as fully wired, clarify WorkerRegistration remains unwired - PRODUCTION_PATH_CHANGELOG.md: add heartbeat integration changelog with before/after, test inventory, and known gaps Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Prepare acceptance criteria and gap analysis for upcoming WorkerRegistration wiring. No production code changes. - ACCEPTANCE_CRITERIA.md: add G9 gate (8 conditions) for WorkerRegistration production-path integration - WORKER_GAP_ANALYSIS.md: add Section 7 audit covering legacy vs unified type gap, bridge function metadata loss, minimal wiring point, field mapping, and 5 suggested tests - PRODUCTION_PATH_CHANGELOG.md: no changes needed (already accurate for heartbeat wiring only) Key findings: - WorkerRegistrationRead (8 fields) is a strict subset of WorkerRegistration (14 fields) — 6 unified fields missing - Bridge function discards all metadata from process_status - Codex has started implementation (uncommitted) with 5 tests Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Install gstack as a project-local skill set and register available slash commands in CLAUDE.md. Use /browse for all web browsing tasks. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This cleanup consolidates the active housekeeper runtime path onto .
Changes:
Validation:
21 passed in 4.92s
Non-blocking follow-up: