{"full_name":"muratcankoylan/Agent-Skills-for-Context-Engineering","owner":"muratcankoylan","name":"Agent-Skills-for-Context-Engineering","shard":"1","url":"https://github.com/muratcankoylan/Agent-Skills-for-Context-Engineering","description":"A comprehensive collection of Agent Skills for context engineering, multi-agent architectures, and production agent systems. Use when building, optimizing, or debugging agent systems that require effective context management.","homepage":null,"topics":[],"tag_pages":[],"language":"Python","license":"MIT","avatar":"https://avatars.githubusercontent.com/u/132029956?u=9049f843f86c73c27e2c54f3b5d16e263c234570&v=4","created_at":"2025-12-21T02:43:42Z","pushed_at":"2026-10-12T02:06:40Z","metrics":{"stars":18350,"forks":1512,"watchers":98,"issues":23,"prs":115,"releases":3,"issues_open":16,"issues_closed":7,"prs_open":42,"prs_closed":18,"prs_merged":55,"discussions":0},"growth":{"day":{"stars":5,"forks":0,"base_day":"2026-11-15"},"week":{"stars":55,"forks":3,"base_day":"2026-11-04"},"month":{"stars":200,"forks":14,"base_day":"2026-10-17"}},"history":{"dates":["2026-09-06","2026-09-07","2026-09-08","2026-09-09","2026-09-10","2026-09-11","2026-09-12","2026-09-13","2026-09-14","2026-09-15","2026-09-16","2026-09-17","2026-09-18","2026-09-19","2026-09-20","2026-09-21","2026-09-22","2026-09-23","2026-09-24","2026-09-25","2026-09-26","2026-09-27","2026-09-28","2026-09-29","2026-09-30","2026-10-01","2026-10-02","2026-10-03","2026-10-04","2026-10-05","2026-10-06","2026-10-07","2026-10-08","2026-10-09","2026-10-10","2026-10-11","2026-10-12","2026-10-13","2026-10-14","2026-10-15","2026-10-16","2026-10-17","2026-10-18","2026-10-19","2026-10-20","2026-10-21","2026-10-22","2026-10-23","2026-10-24","2026-10-25","2026-10-26","2026-10-27","2026-10-28","2026-10-29","2026-10-30","2026-10-31","2026-11-01","2026-11-02","2026-11-03","2026-11-04","2026-11-10","2026-11-11","2026-11-12","2026-11-13","2026-11-14","2026-11-15","2026-11-16"],"stars":[17922,17932,17936,17947,17952,17955,17959,17963,17967,17976,17989,17994,18003,18010,18010,18011,18013,18016,18019,18028,18030,18036,18039,18041,18049,18052,18054,18059,18066,18072,18077,18083,18087,18094,18094,18101,18104,18113,18122,18132,18146,18150,18152,18167,18174,18178,18185,18189,18197,18202,18215,18221,18228,18238,18245,18252,18261,18269,18287,18295,18327,18337,18341,18343,18345,18345,18350],"forks":[1482,1481,1482,1484,1484,1483,1485,1485,1486,1488,1488,1488,1487,1489,1490,1490,1490,1490,1490,1490,1490,1491,1491,1491,1491,1492,1490,1489,1491,1490,1491,1492,1494,1494,1494,1494,1496,1497,1498,1498,1498,1498,1498,1501,1502,1501,1502,1503,1504,1505,1507,1506,1506,1504,1505,1506,1508,1508,1508,1509,1511,1513,1513,1513,1513,1512,1512],"watchers":[102,103,103,104,104,104,104,104,104,104,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,103,104,104,104,104,104,99,99,99,99,99,98,98,98,98,98,98,null,98,98,98,98,98,98],"issues_open":[null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,16,16,16,16,16,16],"prs_open":[null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,42,42,42,42,42,42]},"boards":{},"same_owner":[],"issues":[{"number":1,"title":"Warehouse cooperation","comments":5,"created_at":"2025-12-22T07:31:52Z"},{"number":35,"title":"Proposal: Add context-benchmarking Skill and CI Example","comments":2,"created_at":"2026-01-20T22:08:57Z"},{"number":108,"title":"Stale researcher-OS baseline numbers in README.md and AGENTS.md","comments":1,"created_at":"2026-07-14T07:53:14Z"},{"number":107,"title":"Research loop fails on Windows due to Unix-only fcntl import","comments":1,"created_at":"2026-07-14T07:51:43Z"},{"number":106,"title":"API loops and judge examples lack cost / rate-limit safeguards","comments":1,"created_at":"2026-07-14T07:50:55Z"},{"number":105,"title":"Hosted-agents Modal example uses shell injection and unsafe secret handling","comments":1,"created_at":"2026-07-14T07:49:37Z"},{"number":104,"title":"Project-development pipeline is vulnerable to path traversal / arbitrary file deletion","comments":1,"created_at":"2026-07-14T07:48:16Z"},{"number":114,"title":"Plugin manifest at .plugin/plugin.json is not read — skills are namespaced by commit SHA","comments":0,"created_at":"2026-07-31T06:04:57Z"}],"discussions":[],"latest_release":{"name":"v2.3.0 — Measured Router Benchmark + Corpus-Wide Skill Hardening","tag":"v2.3.0","published_at":"2026-05-22T04:31:12Z","body":"The first release with measured benchmark results across four frontier models via the Cursor SDK, a corpus-wide hardening pass across all 15 published skills, and the first full-body Stage 3 effectiveness result.\r\n\r\nThe correction in this release is important: we did not stop at the three descriptions that the router benchmark complained about. Every skill body was audited against the same standard, and the machine-readable substrate moved with it: mechanisms, claims, corpus index, activation fixtures, validators, template, docs, PR/release narrative.\r\n\r\n> Pull request: [#87](https://github.com/muratcankoylan/Agent-Skills-for-Context-Engineering/pull/87)\r\n> Full inventory: `CHANGELOG.md`\r\n> Latest router report: `researcher/benchmarks/router/results-published/2026-05-19.md`\r\n> Stage 3 hard pilot: `researcher/benchmarks/effectiveness/results-published/2026-05-19-stage3-hard-pilot.md`\r\n> Project narrative: `researcher/insights/how-we-built-this.md`\r\n> Technical findings: `researcher/insights/auto-research-experiment.md`\r\n> Benchmark plan: `researcher/benchmarks/PLAN.md`\r\n\r\n## Research charts\r\n\r\n![Research-to-Skill Operating System](https://raw.githubusercontent.com/muratcankoylan/Agent-Skills-for-Context-Engineering/agents/autonomous-org-harness/assets/release/v2.3.0/researcher-os-pipeline-v2-3.svg)\r\n\r\n![Router Benchmark Leaderboard](https://raw.githubusercontent.com/muratcankoylan/Agent-Skills-for-Context-Engineering/agents/autonomous-org-harness/assets/release/v2.3.0/router-leaderboard-2026-05-19.svg)\r\n\r\n![Measured Skill Routing Gains](https://raw.githubusercontent.com/muratcankoylan/Agent-Skills-for-Context-Engineering/agents/autonomous-org-harness/assets/release/v2.3.0/skill-routing-gains-v2-3.svg)\r\n\r\n![Corpus-Wide Hardening Metrics](https://raw.githubusercontent.com/muratcankoylan/Agent-Skills-for-Context-Engineering/agents/autonomous-org-harness/assets/release/v2.3.0/corpus-hardening-metrics-v2-3.svg)\r\n\r\n## Stage 2 router results\r\n\r\nThree 600-record sweeps at seed=1, same fixture, same shuffle orders:\r\n\r\n- Baseline\r\n- Post-description rewrite\r\n- Post-corpus hardening\r\n\r\nMeasured routing gains:\r\n\r\n- `context-fundamentals`: 0.255 -> 0.489 top-1 (+23.4pp)\r\n- `project-development`: 0.750 -> 1.000 top-1 (+25pp)\r\n- `tool-design`: 0.729 -> 0.807 top-1 (+7.8pp)\r\n\r\nMay 19 corpus-hardening sweep: 600/600 usable records, 0 format failures after retry.\r\n\r\n| Model | Top-1 | Top-3 |\r\n| --- | --- | --- |\r\n| Gemini 3.1 Pro | 0.920 | 0.933 |\r\n| Composer 2 | 0.913 | 0.947 |\r\n| GPT-5.5 | 0.913 | 0.973 |\r\n| Claude Opus 4.7 | 0.840 | 0.933 |\r\n\r\nRouter negative controls now support `expected_primary_skill: \"none\"`, so basic arithmetic, translation, and generic formatting prompts no longer reward catch-all skill loading.\r\n\r\n## Stage 3 effectiveness result\r\n\r\nThe first Stage 3 pilot was a null result: the runner worked, but the task was too easy. We used that failure correctly and hardened the benchmark.\r\n\r\nThe final hard pilot uses task `002-filesystem-scratchpad-behavior`, which requires a compact `scratch/` evidence artifact from a large diagnostic trace. The runner inlines selected full `SKILL.md` bodies so the benchmark measures actual skill-body content.\r\n\r\nResult on `composer-2`, 3 replications per condition:\r\n\r\n| Condition | Success |\r\n| --- | ---: |\r\n| No-skill control | 0/3 |\r\n| `filesystem-context` target | 3/3 |\r\n| Irrelevant `bdi-mental-states` negative control | 0/3 |\r\n| Target + related skill | 3/3 |\r\n| Target + unrelated skill | 3/3 |\r\n| Full corpus | 1/3 |\r\n\r\nThis is +100pp target-minus-control and +100pp target-minus-negative on a deterministic behavior gate. The full-corpus result is also important: precise activation beat loading every skill, which validates the repository's central context-engineering claim.\r\n\r\n## Corpus-wide hardening\r\n\r\n- All 15 skill bodies now include explicit ownership, adjacent `Do not activate` routing, practical guidance, examples, gotchas, integration boundaries, and references.\r\n- `bdi-mental-states","is_prerelease":false},"readme":true,"readme_zh":false,"news":[],"stale":false,"snapshot_at":"2026-11-16T09:23:14Z","history_start":"2026-09-06"}