{"id":50735921,"url":"https://github.com/p10ns11y/skills","last_synced_at":"2026-06-10T13:30:44.377Z","repository":{"id":363287285,"uuid":"1262661543","full_name":"p10ns11y/skills","owner":"p10ns11y","description":"Skills copied from my personal projects","archived":false,"fork":false,"pushed_at":"2026-06-08T08:12:07.000Z","size":129,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"master","last_synced_at":"2026-06-08T09:25:22.538Z","etag":null,"topics":["agents","ai-coding","ai-optimization","ai-orchestration","ai-security","ai-tools","high-signal","skills"],"latest_commit_sha":null,"homepage":"","language":"Shell","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/p10ns11y.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":"audit-allow-builds/SKILL.md","citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-06-08T07:44:54.000Z","updated_at":"2026-06-08T08:24:43.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/p10ns11y/skills","commit_stats":null,"previous_names":["p10ns11y/skills"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/p10ns11y/skills","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/p10ns11y%2Fskills","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/p10ns11y%2Fskills/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/p10ns11y%2Fskills/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/p10ns11y%2Fskills/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/p10ns11y","download_url":"https://codeload.github.com/p10ns11y/skills/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/p10ns11y%2Fskills/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":34155422,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-06-10T02:00:07.152Z","response_time":89,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agents","ai-coding","ai-optimization","ai-orchestration","ai-security","ai-tools","high-signal","skills"],"created_at":"2026-06-10T13:30:43.397Z","updated_at":"2026-06-10T13:30:44.364Z","avatar_url":"https://github.com/p10ns11y.png","language":"Shell","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Agent Skills Library\n\n**After losing and winning many battles — and winning the war — these are the playbooks that survived.**\n\nNot blog-post advice. Not \"best practices\" copied from docs. Each skill here is a scar from something that actually broke: agents that overwrote each other's work, context windows that ate the budget before the fix landed, a CV promote that almost hit production, supply-chain installs that looked clean until they weren't. The wins got encoded. The losses became guardrails.\n\nThis is a library of [Cursor Agent Skills](https://cursor.com/docs/context/skills) (also usable with Grok, Continue, and other agentic tools) for **token-efficient context**, **safe multi-agent orchestration**, **supply-chain hardening**, and **self-guarded autonomous workflows**.\n\nEach directory is a self-contained skill with a `SKILL.md` frontmatter (`name`, `description`) that agents use for discovery and routing.\n\n## Why these skills exist\n\nMost agent failures are not model failures. They are **context**, **coordination**, and **guardrail** failures — the kind you only recognize after you've paid for them once.\n\n\n| Problem | What we lost | What the skill encodes |\n| ------- | ------------ | ---------------------- |\n| Context window bloat on large repos | Hours re-explaining the same codebase every turn | [ai-optimization](ai-optimization/SKILL.md) — relevance scoring, compression, strict budgets |\n| One-shot help that doesn't compound | Same architectural mistakes, session after session | [fusion-sage](fusion-sage/SKILL.md) — synthesis + surplus (Q \u003e 1 improvements) |\n| Parallel agents overwriting each other | Merged garbage, lost work, `cp` disasters | [git-worktrees](git-worktrees/SKILL.md) + [agent-orchestrator](agent-orchestrator/SKILL.md) |\n| Unsafe autonomous actions | Near-misses on CV writes, runaway API spend | [finder-reactor](finder-reactor/SKILL.md) + [cv-promote-guard](cv-promote-guard/SKILL.md) |\n| npm supply-chain risk | \"Audit clean\" installs that still weren't safe | [fix-dependency-security](fix-dependency-security/SKILL.md) + [supply-chain-harden](supply-chain-harden/SKILL.md) |\n\n**Philosophy:** Fission keeps you fast. Fusion makes you unstoppable. Guards are what you add after you almost didn't need them.\n\n---\n\n## Quick start\n\n### 1. Clone\n\n```bash\ngit clone \u003cthis-repo\u003e ~/skills   # or fork and clone yours\ncd ~/skills\n```\n\n### 2. Install for Cursor\n\n**Global** (available in every project):\n\n```bash\nmkdir -p ~/.cursor/skills\nfor skill in ai-optimization fusion-sage agent-orchestrator git-worktrees; do\n  ln -sf \"$(pwd)/$skill\" ~/.cursor/skills/$skill\ndone\n```\n\n**Per-project** (share with teammates via repo):\n\n```bash\nmkdir -p .cursor/skills\nln -sf /path/to/skills/ai-optimization .cursor/skills/ai-optimization\n```\n\nCursor discovers skills from `~/.cursor/skills/` and `.cursor/skills/`. The agent loads them when the task matches the skill's `description`.\n\n### 3. Install for Grok / other agents\n\n```bash\nmkdir -p ~/.grok/skills\nln -sf \"$(pwd)/ai-optimization\" ~/.grok/skills/ai-optimization\n# repeat for skills you need\n```\n\nPoint your project's `AGENTS.md` (or equivalent) at the active skill set and verification commands.\n\n### 4. Optional — project overlays\n\nSkills are **portable by default**. Repo-specific tuning (relevance scoring, fusion targets, devcontainer starters) lives in [examples/](examples/README.md) — copy overlays into your project rather than editing skill bodies.\n\n---\n\n## Recommended starter packs\n\nPick a pack, symlink those skills, add more as needed.\n\n\n| Pack                            | Skills                                                                                        | Best for                           |\n| ------------------------------- | --------------------------------------------------------------------------------------------- | ---------------------------------- |\n| **Context \u0026 speed**             | `ai-optimization`, `fusion-sage`                                                              | Large codebases, expensive context |\n| **Multi-agent delivery**        | `agent-orchestrator`, `git-worktrees`, `concurrent-cli-agents`, `split-to-prs`                | Parallel agents, PR hygiene        |\n| **Node supply chain**           | `fix-dependency-security`, `upgrade-packages`, `audit-allow-builds`, `supply-chain-harden`    | pnpm monorepos, CI hardening       |\n| **React / Next frontend**       | `react-client-expert`, `semantic-markup-css`, `project-editor-profile`                        | App Router, a11y, editor sync      |\n| **Opportunity finder platform** | `finder-reactor`, `agentic-reactor`, `x-agent-resources`, `cv-promote-guard`, `tauri-agentic` | Tauri + X + LLM autonomous loops   |\n| **C / CMake / MVU**             | `explore-repo-readonly`, `mvu-refactor-plan`, `src-tree-reorganize`                           | elomaxz-style refactors            |\n| **Strategy \u0026 decisions**        | `higher-order-decision-architect`, `bdd-strategizer`                                          | Architecture tradeoffs, pre-mortem, test-first refactors |\n\n\n---\n\n## Skill catalog\n\n### Agent orchestration \u0026 execution\n\n\n| Skill                                                       | One-liner                                                                                          |\n| ----------------------------------------------------------- | -------------------------------------------------------------------------------------------------- |\n| [agent-orchestrator](agent-orchestrator/SKILL.md)           | Triage single-shot vs full orchestration; briefs, worktrees, independent verification, clean merge |\n| [concurrent-cli-agents](concurrent-cli-agents/SKILL.md)     | Run multiple CLI agents safely on isolated worktrees or sandboxes                                  |\n| [git-worktrees](git-worktrees/SKILL.md)                     | Safe git worktree usage for agents; commit-then-merge; disk hygiene                                |\n| [gt-flow](gt-flow/SKILL.md)                                 | When to use Graphite (`gt`) vs plain branches with agent worktrees                                 |\n| [split-to-prs](split-to-prs/SKILL.md)                       | Split a body of work into small, reviewable PRs                                                    |\n| [subagent-delegation](subagent-delegation/SKILL.md)         | Delegate readonly exploration with a strict return format                                          |\n| [subagent-explore-report](subagent-explore-report/SKILL.md) | Structured readonly repo survey (CMake/MVU, etc.)                                                  |\n\n\n### Context, intelligence \u0026 reactors\n\n\n| Skill                                       | One-liner                                                                               |\n| ------------------------------------------- | --------------------------------------------------------------------------------------- |\n| [ai-optimization](ai-optimization/SKILL.md) | Fission engine: relevance scoring, compression, token budgets, progressive disclosure   |\n| [fusion-sage](fusion-sage/SKILL.md)         | Fusion on fission: synthesis, surplus (Q\u003e1), knowledge graph, self-amplification        |\n| [finder-reactor](finder-reactor/SKILL.md)   | Self-guarded opportunity engine: search → analyze → prep → promote with guards \u0026 pauses |\n| [agentic-reactor](agentic-reactor/SKILL.md) | Glue patterns for pause-aware autonomous platforms (finder + shell + MCP)               |\n\n\n### Platform, desktop \u0026 tooling\n\n\n| Skill                                                     | One-liner                                                                      |\n| --------------------------------------------------------- | ------------------------------------------------------------------------------ |\n| [tauri-agentic](tauri-agentic/SKILL.md)                   | Agentic Tauri apps: MCP tools, command palette, guards in Rust + React         |\n| [tauri-ipc-debug](tauri-ipc-debug/SKILL.md)               | Systematic Tauri IPC triage (MVU → invoke → Rust → storage/API)                |\n| [devcontainer-hardened](devcontainer-hardened/SKILL.md)   | Minimal hardened `.devcontainer` configs (pinned digests, non-root, cap drops) |\n| [project-editor-profile](project-editor-profile/SKILL.md) | Generate VS Code/Cursor settings from `.editor/profile.json`                   |\n| [chrome-extension-mv3](chrome-extension-mv3/SKILL.md)     | MV3 extensions: service worker, message passing, content scripts, permissions |\n\n\n### Frontend \u0026 UI\n\n\n| Skill                                               | One-liner                                                                   |\n| --------------------------------------------------- | --------------------------------------------------------------------------- |\n| [react-client-expert](react-client-expert/SKILL.md) | Senior React client patterns: minimal state, Suspense + Query, XState, a11y |\n| [semantic-markup-css](semantic-markup-css/SKILL.md) | Semantics, ARIA, contrast, native elements + Tailwind where CSS wins        |\n\n\n### X / social data \u0026 profile safety\n\n\n| Skill                                           | One-liner                                                                     |\n| ----------------------------------------------- | ----------------------------------------------------------------------------- |\n| [x-agent-resources](x-agent-resources/SKILL.md) | Official X Developer Platform primitives (llms.txt, skill.md, MCP, xurl)      |\n| [cv-promote-guard](cv-promote-guard/SKILL.md)   | Sidecar-first, auditable guards for CV read/promote to external profile repos |\n\n\n### Maintenance, security \u0026 supply chain\n\n\n| Skill                                                       | One-liner                                                 |\n| ----------------------------------------------------------- | --------------------------------------------------------- |\n| [supply-chain-harden](supply-chain-harden/SKILL.md)         | pnpm workspace hardening, audits, Socket Firewall (`sfw`) |\n| [fix-dependency-security](fix-dependency-security/SKILL.md) | Fix vulns/deprecations with safe upgrade patterns         |\n| [upgrade-packages](upgrade-packages/SKILL.md)               | Safe major/minor/patch upgrades with codemods             |\n| [audit-allow-builds](audit-allow-builds/SKILL.md)           | Audit `allowBuilds` / lifecycle scripts in pnpm lockfile  |\n| [audit-ide-dependencies](audit-ide-dependencies/SKILL.md)   | Audit Cursor/VS Code extensions and bundled deps          |\n\n\n### Strategy \u0026 authoring\n\n\n| Skill                                                   | One-liner                                            |\n| ------------------------------------------------------- | ---------------------------------------------------- |\n| [bdd-strategizer](bdd-strategizer/SKILL.md)             | Core-first BDD/TDD decomposition for large refactors |\n| [author-workflow-skill](author-workflow-skill/SKILL.md) | Author new well-formed `SKILL.md` files              |\n| [higher-order-decision-architect](higher-order-decision-architect/SKILL.md) | First-principles decision framework before material architecture/API/security choices |\n\n\n### Specialized\n\n\n| Skill                                                           | One-liner                                              |\n| --------------------------------------------------------------- | ------------------------------------------------------ |\n| [cursor-transcript-harvest](cursor-transcript-harvest/SKILL.md) | Harvest Cursor sessions for prompt tuning / eval data  |\n| [explore-repo-readonly](explore-repo-readonly/SKILL.md)         | Safe pre-change exploration of complex codebases       |\n| [mvu-refactor-plan](mvu-refactor-plan/SKILL.md)                 | CMake + elomaxz MVU refactor planning and execution    |\n| [src-tree-reorganize](src-tree-reorganize/SKILL.md)             | Move scattered sources under `src/` with build updates |\n\n\n---\n\n## Where these were forged\n\nThese skills weren't designed in a vacuum. They were **extracted from real projects after the fighting stopped** — [devprofile](https://github.com/p10ns11y/devprofile), [collab-finder](https://github.com/p10ns11y/collab-finder), [premflow](https://github.com/thecuriousts/premflow), [sorkalam-extension](https://github.com/p10ns11y/sorkalam-extension), and the harvest pipeline in [agent-prompt-tuning-lab](https://github.com/p10ns11y/agent-prompt-tuning-lab) — each with its own stack, its own failures, its own fixes that held up under pressure.\n\nProject names and repo-specific paths live in [examples/README.md](examples/README.md), not in skill bodies. You get the portable lesson; the war stories stay in the footnotes.\n\n\n| Project           | Stack                                                              | Primary skills                                                      |\n| ----------------- | ------------------------------------------------------------------ | ------------------------------------------------------------------- |\n| [**devprofile**](https://github.com/p10ns11y/devprofile)    | Next.js 16, Biome, Tailwind v4, host-browser E2E, CV/PDF, X post search picker, xAI API, xAI collections (embedded datasets) | Context, fusion, frontend, devcontainer, supply chain               |\n| [**collab-finder**](https://github.com/p10ns11y/collab-finder) | Tauri 2, Rust reactor, React, X API + LLM, MCP, SQLite             | Finder reactor, Tauri agentic, X resources, CV guard, git worktrees |\n| [**premflow**](https://github.com/thecuriousts/premflow)      | C, CMake, elomaxz MVU                                              | MVU refactor, readonly explore, src reorganize                      |\n| [**sorkalam-extension**](https://github.com/p10ns11y/sorkalam-extension) | MV3 Chrome extension, vanilla JS, Wiktionary + Tamil VU glossary   | chrome-extension-mv3                                                |\n| [**thepulimaangani**](https://github.com/p10ns11y/thepulimaangani) | Tamil prosody, WASM parser, metre prediction | higher-order-decision-architect |\n| [**agent-prompt-tuning-lab**](https://github.com/p10ns11y/agent-prompt-tuning-lab) | Local Cursor transcript harvest → normalize → split → distill rules/skills | cursor-transcript-harvest, author-workflow-skill |\n\n\nCopy overlays from [examples/overlays/](examples/overlays/) when adapting skills to a similar stack.\n\n---\n\n## Using skills in your project\n\n1. **Symlink** the skills you need (global or `.cursor/skills/`).\n2. **Declare** active skills in `AGENTS.md` with verification commands (`pnpm type-check`, `pnpm lint`, etc.).\n3. **Optional:** Add a Cursor rule (`.cursor/rules/*.mdc`) to always apply context skills on large repos.\n4. **Optional:** Copy a project overlay from [examples/](examples/README.md) into `.agents/skills/\u003cskill\u003e/references/`.\n\n### Skill anatomy\n\n```\nmy-skill/\n├── SKILL.md          # Required — frontmatter + instructions\n├── reference.md      # Optional deep docs\n├── scripts/          # Optional utilities agents can run\n└── templates/        # Optional briefs, configs\n```\n\nDescriptions must be **third-person**, **specific**, and include **trigger terms** — agents route on `description`, not the title.\n\nAuthor new skills with [author-workflow-skill](author-workflow-skill/SKILL.md).\n\n### Optional Cursor rules (`rules/`)\n\nSome skills ship a thin **`.mdc` router** in [`rules/`](rules/) for `alwaysApply: true` behavior — e.g. [higher-order-decision-architect.mdc](rules/higher-order-decision-architect.mdc) loads the full skill on material decisions without duplicating the framework in every project. Symlink into `.cursor/rules/` alongside skills.\n\n---\n\n## Contributing\n\nIf you've lost a battle worth documenting, contribute the win:\n\n1. Keep `SKILL.md` under ~500 lines; put depth in `reference.md` or `examples/`.\n2. **No project-specific paths in skill bodies** — use [examples/](examples/README.md) for overlays and provenance.\n3. Prefer linking between skills over duplicating guardrails.\n4. Include concrete verification steps and anti-patterns — especially the ones that cost you a day.\n\n---\n\n## License\n\n[MIT](LICENSE) — use, copy, modify, and ship these skills however you want. Keep the copyright notice if you redistribute the files.\n\nIf a skill saves you a battle, a post on [X (@peramanathan)](https://x.com/peramanathan) helps others find it. No obligation — just good karma if you're inclined.\n\n**You don't need to lose the same battles we did.**\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fp10ns11y%2Fskills","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fp10ns11y%2Fskills","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fp10ns11y%2Fskills/lists"}