{"id":50119578,"url":"https://github.com/naoray/skills","last_synced_at":"2026-05-23T18:04:45.389Z","repository":{"id":351857527,"uuid":"1212800091","full_name":"Naoray/skills","owner":"Naoray","description":"Naoray's personal Claude Code, Cursor \u0026 Codex skills. Connect in one command with scribe.","archived":false,"fork":false,"pushed_at":"2026-05-19T08:10:53.000Z","size":342,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-05-19T09:14:11.427Z","etag":null,"topics":["ai-agents","claude-code","codex","cursor","scribe","skills"],"latest_commit_sha":null,"homepage":"https://github.com/Naoray/scribe","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/Naoray.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"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":"2026-04-16T18:40:57.000Z","updated_at":"2026-05-19T08:10:57.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/Naoray/skills","commit_stats":null,"previous_names":["naoray/skills"],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/Naoray/skills","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Naoray%2Fskills","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Naoray%2Fskills/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Naoray%2Fskills/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Naoray%2Fskills/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/Naoray","download_url":"https://codeload.github.com/Naoray/skills/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Naoray%2Fskills/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":33406497,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-23T04:15:53.637Z","status":"ssl_error","status_checked_at":"2026-05-23T04:15:53.242Z","response_time":53,"last_error":"SSL_read: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"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":["ai-agents","claude-code","codex","cursor","scribe","skills"],"created_at":"2026-05-23T18:04:19.519Z","updated_at":"2026-05-23T18:04:45.357Z","avatar_url":"https://github.com/Naoray.png","language":"Shell","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Naoray/skills\n\n**15 production-grade skills for Claude Code, Codex, and Gemini.**\n Each one earned its slot against a deterministic 15-check audit, ships with router/structure evals, and declares its evidence tier upfront — so you know whether it's empirical, practitioner-backed, or heuristic before you install.\n\nInstall one. Install the catalog. Built for [Scribe](https://github.com/Naoray/scribe), works without it.\n\n```bash\nscribe registry connect Naoray/skills\n```\n\n## Kits — install in bundles, not skill-by-skill\n\nFive stackable kits cover the catalog. Each maps to one theme — pick the kits that match your work, layer multiple if you do more than one thing.\n\n| Kit | Skills | Use it for |\n|---|---|---|\n| **`daily-workflow`** | `plan-my-day`, `evaluate-day`, `session-plan`, `record`, `meeting` | Plan, capture, and close the day from inside the agent. |\n| **`release-pipeline`** | `changelog-pr`, `code-review-artifacts`, `visual-review`, `cleanup` | PR bodies, comprehension diagrams, rendered-output verification, post-ship hygiene. |\n| **`orchestration`** | `orchestrator-mode`, `orchestrator-handoff` | Multi-agent coordination over [Solo MCP](https://github.com/sublayerapp/solo); state handoff between sessions. |\n| **`methodology`** | `research-mode`, `skill-creator` | Anti-hallucination citation discipline; author your own skills with the 5-part trigger contract. |\n| **`mac-productivity`** | `apple-calendar`, `dev-browser` | Calendar.app CRUD via AppleScript + persistent-page browser automation. |\n\nReference each kit in your project's `.scribe.yaml`:\n\n```yaml\nkits:\n  - daily-workflow\n  - release-pipeline\n```\n\nThen `scribe sync`. Kits stack; add or remove individual skills on top with `add:` / `remove:`. Kit bodies live in [`kits/`](kits/) and are surfaced via `scribe.yaml`.\n\n## What's inside\n\n### Daily workflow (kit: `daily-workflow`)\n\n- **`plan-my-day`** — Fuse your calendar, reminders, and intent into a reviewed time-block schedule and a daily-note scaffold. Asks before it touches the calendar.\n- **`evaluate-day`** — Turn today's Running Log into a grounded end-of-day reflection inside the daily note. Won't overwrite without confirmation.\n- **`session-plan`** — A focused practice/work block with one main outcome, timed sub-blocks, and an explicit \"ignore this today\" list. Reads persistent domain context from `~/Context/\u003cdomain\u003e/`.\n- **`record`** — `/record \u003cmessage\u003e` appends one deduplicated, timestamped entry to today's Running Log. Used by the other skills as their event-capture primitive.\n- **`meeting`** — Live note capture while a meeting is active. `/meeting \u003ctitle\u003e` opens a session; every later message becomes raw notes, decisions, and action items until `/meeting end`.\n\n### Release pipeline (kit: `release-pipeline`)\n\n- **`changelog-pr`** — Generate or refresh a release PR body and `CHANGELOG.md` in Keep a Changelog format. Detects squash vs merge style; emits only the sections that have entries.\n- **`code-review-artifacts`** — Diff briefs, ASCII flow maps, architecture x-rays. Comprehension artifacts for reviewers and AI agents about to touch unfamiliar code. Every artifact carries an explicit \"Inspected / Not inspected\" scope line.\n- **`visual-review`** — Rendered-output verification across browser pages, PRs, CLI commands, and TUIs. Captures screenshots/recordings plus a visible-issues report. Read-only — never commits.\n- **`cleanup`** — Post-ship project hygiene. Classifies artifacts as remove / update / keep / review, applies changes only after you confirm. Not a disk-space tool.\n\n### macOS productivity (kit: `mac-productivity`)\n\n- **`apple-calendar`** — Calendar.app CRUD on macOS: list, create, update, delete, search; handles recurrence safely. Direct AppleScript — no Google/Outlook bridging.\n- **`dev-browser`** — Persistent-page browser automation for QA, scraping, and login flows. Headless Chromium with state survival across commands. Refuses destructive or payment actions without explicit consent.\n\n### Multi-agent orchestration (kit: `orchestration`)\n\n- **`orchestrator-mode`** — Convert the current session into a delegating coordinator over [Solo MCP](https://github.com/sublayerapp/solo). Sets agent-selection rules (codex for coding, claude for skills/slash-commands, gemini for second-opinion \u0026 dissent), worktree-by-default isolation, and scratchpad-based feedback capture.\n- **`orchestrator-handoff`** — A paste-ready prompt for the next orchestrator session. Captures in-flight agents, scratchpads, locked decisions, open PRs, and dispatch intent so the next window starts hot.\n\n### Methodology (kit: `methodology`)\n\n- **`research-mode`** — Anti-hallucination mode: require citations, surface conflicts, refuse to present uncertain claims as fact. Toggleable; built for spec review and source-grounded analysis.\n- **`skill-creator`** — Author or revise an AI-agent skill using the 5-part trigger contract, evidence-tier (E / P / H) gating, and progressive disclosure. LLM-agnostic and registry-agnostic (scribe, `.claude/skills`, `.ai/skills`, custom backends). Mandatory reviewer pass before declaring done.\n\n## Why these skills don't misfire\n\nEvery skill in this catalog ships with two things most don't:\n\n**A 5-part `description` contract.** Use when / Inputs / Do not use when / Produces / Escalate if. The router only sees the description on every invocation, so the contract is the trigger boundary. No \"always use this\" pushy phrasing. Sharp disqualifiers prevent near-miss firing.\n\n**Routing + structural evals.** `evals/trigger.csv` replays real prompts (positives that should fire + near-misses that share keywords but shouldn't) against the router. `evals/checks.md` inherits 15 deterministic structural checks (frontmatter, body length, provenance, no pushy language, etc.). Description edits and model upgrades re-run the regression — broken triggers fail before they merge.\n\nRead [`skill-creator/SKILL.md`](skill-creator/SKILL.md) for the full procedure and [`skill-creator/evals/checks.md`](skill-creator/evals/checks.md) for the 15-check audit.\n\n## How progressive disclosure pays off\n\nLarger skills are folders, not monoliths:\n\n```text\nskill-name/\n  SKILL.md          # small router — trigger contract + evidence tier + entry points\n  workflows/        # loaded only when a specific mode fires\n  references/       # models, rubrics, principles loaded on demand\n  scripts/          # deterministic helpers (no LLM needed)\n  evals/            # trigger.csv + checks.md — skill tests, not runtime context\n```\n\nThe router stays under ~120 lines, so the LLM sees only the contract on every invocation. Mode-specific bodies load when triggered. Evals never enter runtime context — they live on disk for regression replay.\n\n## Install\n\n### Via Scribe (recommended)\n\nConnect the registry once:\n\n```bash\nscribe registry connect Naoray/skills\n```\n\nThen install by kit, by skill, or browse:\n\n```bash\n# Kit (recommended — bundles for a use case)\n# Declare kits in your project's .scribe.yaml under `kits:`, then:\nscribe sync\n\n# Single skill (interactive picker)\nscribe browse\n\n# Or by name (legacy add, still works)\nscribe add Naoray/skills:\u003cskill-name\u003e\n```\n\nScribe keeps the canonical copy in `~/.scribe/skills/` and links it into Claude Code, Codex, and Gemini.\n\nWant all 15 skills at once?\n\n```bash\nscribe install --all\n```\n\nInstalls every catalog entry from connected registries in one shot.\n\n### Let your AI set it up\n\nPaste this into Claude Code, Codex, or any agentic LLM — works whether scribe is installed or not:\n\n```\nHelp me set up the Naoray/skills registry (https://github.com/Naoray/skills) for my AI coding tools.\n\n1. Check if scribe is installed: `scribe --version`. If missing, install from\n   https://github.com/Naoray/scribe and make sure `gh auth status` succeeds.\n\n2. Connect the registry: `scribe registry connect Naoray/skills`.\n\n3. Ask whether I want a kit, cherry-pick skills, or the whole catalog:\n   - Kit (recommended): show the five kits (daily-workflow, release-pipeline,\n     orchestration, methodology, mac-productivity), add my picks to the project's\n     `.scribe.yaml` under `kits:`, then run `scribe sync`.\n   - Cherry-pick: `scribe browse` (interactive picker) or `scribe add Naoray/skills:\u003cname\u003e`.\n   - Whole catalog: `scribe install --all`.\n\n4. Confirm final state with `scribe list`.\n\nPause before any install or connect command so I can approve.\n```\n\n### Manual (no Scribe)\n\nClone and symlink. Example for Claude Code:\n\n```bash\ngit clone https://github.com/Naoray/skills.git ~/skills-src\nln -s ~/skills-src/plan-my-day   ~/.claude/skills/plan-my-day\nln -s ~/skills-src/skill-creator ~/.claude/skills/skill-creator\n```\n\n## Stay in sync\n\n```bash\nscribe sync\n```\n\nUpdates only the skills you already installed. Never adds new ones behind your back.\n\n## Curation policy\n\nEvery skill in `scribe.yaml` cleared the [`skill-creator`](skill-creator/SKILL.md) fit gate: repeated workflow (2+ prior real invocations), stable artifact, durable cognitive model, sharp trigger boundary, unique slot, declared evidence tier, acceptable overreach risk.\n\nWhat's intentionally **not** in the catalog:\n\n- **`recap`, `recap-report`, `scribe-agent`** — recap is being rethought; scribe ships its own agent skill.\n\nPast consolidations:\n\n- `code-review-artifacts` replaces `flow-review` and `xray`.\n- `visual-review` covers both browser UI and CLI/TUI review.\n- `changelog-pr` covers release PR bodies and `CHANGELOG.md` updates.\n\n## Repo layout\n\n```text\n.\n├── plan-my-day/           # daily-workflow kit\n├── evaluate-day/\n├── session-plan/\n├── record/\n├── meeting/\n├── changelog-pr/          # release-pipeline kit\n├── code-review-artifacts/\n├── visual-review/\n├── cleanup/\n├── orchestrator-mode/     # orchestration kit\n├── orchestrator-handoff/\n├── research-mode/         # methodology kit\n├── skill-creator/\n├── apple-calendar/        # mac-productivity kit\n├── dev-browser/\n├── kits/                  # kit manifests\n│   ├── daily-workflow.yaml\n│   ├── release-pipeline.yaml\n│   ├── orchestration.yaml\n│   ├── methodology.yaml\n│   └── mac-productivity.yaml\n└── scribe.yaml            # registry manifest (publishes catalog + kits)\n```\n\nEach skill is self-contained — install one without the others, or pull in a kit.\n\n## Contributing\n\nIssues and PRs welcome. New skill proposals must pass the [`skill-creator`](skill-creator/SKILL.md) fit gate; open an issue with the proposed 5-part trigger contract and evidence tier before writing the SKILL.md.\n\n## New to scribe?\n\nStart here → [Naoray/scribe](https://github.com/Naoray/scribe)\n\n## License\n\nSee `LICENSE`.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fnaoray%2Fskills","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fnaoray%2Fskills","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fnaoray%2Fskills/lists"}