| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
feat: v1.1.0 executor mode — zero-intervention delivery, detail catalog, implementation audit - SKILL.md: Executor Mode entry (SPEC/PLAN/BUILD/AUDIT/REGRESS), zero-intervention iron rules, default assumptions, 25/26/27 routing, delivery gates - references/25-autonomous-delivery.md: 12-dimension spec completion, assumption ledger, vertical slices, delivery state machine, checkpoint resume, degradation matrix, delivery report - references/26-product-detail-catalog.md: surface decision table, composer (search toggle, IME guard, attachments), message actions (copy/delete/undo/regenerate), session list, four states, keyboard, responsive - references/27-implementation-audit.md: requirement traceability matrix, five-hop wiring, fake-implementation detection, evidence ledger, three-round audit - tools/audit-skill.mjs: static audit (TODO/empty handler/console.log/mock/secrets/routing/claims) with selftest and --root project mode; npm run verify chains check+audit+selftest - fix drift: ghost CI references in CHANGELOG, line-count claims (21k actual vs 31k claimed), stale section names, Node >=20.11 unified; install docs added (README x3 + QUICK_START §0); site updated to 27 nodes | 6 天前 | |
feat: expand all references with 2026 mature-agent-stack details + ecosystem research - Add docs/TECH_LANDSCAPE.md: 2022-2026 mature tech landscape (agent clusters, MCP stateless spec, A2A, memory/RAG, observability/eval/security, big-tech production patterns, kill list) - Add docs/research/: five sourced reports (orchestration, MCP, big-tech, memory/RAG, observability/eval/security) - Expand all 24 references (01-24) with deeper engineering rationale, comparison tables, edge cases, and integration examples using mature solutions - Incorporate agent-cluster (orchestrator-worker/handoffs/A2A), MCP Streamable HTTP + OAuth 2.1, hybrid search + rerank, Mem0/Zep/LangMem, promptfoo/LLM-as-judge, Langfuse/OTel semconv, LiteLLM gateway, Upstash Ratelimit, Stripe idempotent webhooks, BullMQ, OWASP LLM Top 10 - Update REUSE_MATRIX mappings; add apps/cli to allowed layout prefixes - Mark deprecated tech (Swarm, AutoGen legacy, Assistants API, NeMo Guardrails, MCP HTTP+SSE) explicitly - Gate: npm run check 0 failures, 0 notes | 24 天前 | |
fix: split offline verify from runtime probe (no false green) - npm run verify = check + audit + audit:selftest (offline, no network needed) - npm run verify:full = verify + probe (requires .probe-runtime install) - probe prints explicit SKIP when scratch deps absent, and no longer advertises a nonexistent --install flag - rationale: a gate that silently skips must not be reported as passing | 5 天前 |