Official description2 sourcesOpen source5d ago
Ottili-ONE/skills
Claude Code, Codex, OpenClaw and Ottili AI Skills - powered by Ottili ONE
View on GitHub ↗Direct link · no redirect page
- Star count
- 1 ★ · Ottili-ONE/skills · read from the GitHub search API · (checked 2026-10-16 20:00:20)
- Source
- read directly from its GitHub page · type: open-source project
- Harvest history
- first seen 2026-10-12 · last seen 2026-10-16 · seen 7× · listed by 2 sources
- Score
- score 203 · 1 ★ · 2 sources · repo
FAQ (answers come from our records)
How do I install it?
We did not find an install command for this repository (our install coverage is only about 4%). Check its README or releases page.
Who is it for?
We filed it under these tags: 开源项目. The tags come from the entry itself, not a taxonomy we imposed.
Who maintains it?
repository
Ottili-ONE/skills；first catalogued 2026-10-12；last seen 2026-10-16。There is no official certification or quality rating here — we record sources and dates only.Third-party quality signals
Ottili ONE Skills
以下摘录自该仓库的 README（片段） · 在 GitHub 读完整 README →
Agent Skills for Claude Code, Codex, Cursor, OpenClaw, Ottili AI and Ottili Coder.
Skills are folders of instructions, scripts, and resources that AI agents discover and apply to
specific tasks. This repository is the single source of truth for those skills: it hosts the
skill folders, the catalog that describes them, the tooling that validates and installs them,
and the engineering playbooks (skills/engineering/**) that Ottili agents use to build reliable
software.
Repository layout
Installing skills into an agent
Each agent looks for skills in its own directory. The convention for every tool below is: a skill
is a folder whose name is the skill's name, containing a SKILL.md with YAML frontmatter.
Copying (or symlinking) the folder into the agent's skills directory is enough; restart the
agent to load it.
Verified skill-directory conventions (retrieved 2026-10-07):
- **Claude Code** — ~/.claude/skills/ /SKILL.md. Skills are auto-discovered from the
skills directory; --settings and CLAUDE.md can scope which skills load.
Source: (HTTP 301 -> 200).
- **Codex (OpenAI)** — $CODEX_HOME/skills/ /SKILL.md, typically ~/.codex/skills/.
Skills are loaded from this directory and invoked by name.
Source: (HTTP 200).
- **Cursor** — ~/.cursor/skills/ /SKILL.md (project-scoped .cursor/skills/ also works).
Source: (HTTP 308 redirect).
- **OpenClaw / generic** — ~/.agents/skills/ /SKILL.md or any directory the runtime
announces; the layout is the same folder + SKILL.md convention.
- **Anthropic skills collection** — (HTTP 200) ships
reusable skill folders in exactly this layout; use it as a reference for what a well-formed
skill looks like.
scripts/gen-catalog.mjs
catalog.json is **generated**, never hand-edited. scripts/gen-catalog.mjs walks
every skills/ / /SKILL.md, reads the YAML frontmatter and writes one
entry per skill. Packages are **derived from the skills/ folders** at generation time —
a package is any top-level folder under skills/ that contains skill subfolders — so
there is no hand-maintained package or skill list.
CI runs node scripts/gen-catalog.mjs --check on every push and PR
(.github/workflows/validate.yml), so a hand-edited catalog.json cannot land.
scripts/install.sh
--target accepts codex, claude, cursor, agents, or an explicit directory path.
Without --force, an existing target folder is left untouched and the script exits non-zero.
--dry-run prints the source and destination paths it would touch and exits without writing.
Packages are **derived from the skills/ folders at install time** — never a
hand-maintained list and never read from catalog.json. A package is any top-level
folder under skills/ that contains at least one subfolder with a SKILL.md, so a
new package or skill appears automatically. --package installs every skill in
that package:
Virtual packages: generic installs every skill; meta and ref-* are reserved
on-demand names that currently resolve to the empty set. The source path for each
skill is resolved by searching the skills/ tree rooted at the script's own
directory, so the folder name alone is enough — no catalog lookup.
scripts/validate.py
Skills are folders of instructions, scripts, and resources that AI agents discover and apply to
specific tasks. This repository is the single source of truth for those skills: it hosts the
skill folders, the catalog that describes them, the tooling that validates and installs them,
and the engineering playbooks (skills/engineering/**) that Ottili agents use to build reliable
software.
Repository layout
Installing skills into an agent
Each agent looks for skills in its own directory. The convention for every tool below is: a skill
is a folder whose name is the skill's name, containing a SKILL.md with YAML frontmatter.
Copying (or symlinking) the folder into the agent's skills directory is enough; restart the
agent to load it.
Verified skill-directory conventions (retrieved 2026-10-07):
- **Claude Code** — ~/.claude/skills/ /SKILL.md. Skills are auto-discovered from the
skills directory; --settings and CLAUDE.md can scope which skills load.
Source: (HTTP 301 -> 200).
- **Codex (OpenAI)** — $CODEX_HOME/skills/ /SKILL.md, typically ~/.codex/skills/.
Skills are loaded from this directory and invoked by name.
Source: (HTTP 200).
- **Cursor** — ~/.cursor/skills/ /SKILL.md (project-scoped .cursor/skills/ also works).
Source: (HTTP 308 redirect).
- **OpenClaw / generic** — ~/.agents/skills/ /SKILL.md or any directory the runtime
announces; the layout is the same folder + SKILL.md convention.
- **Anthropic skills collection** — (HTTP 200) ships
reusable skill folders in exactly this layout; use it as a reference for what a well-formed
skill looks like.
scripts/gen-catalog.mjs
catalog.json is **generated**, never hand-edited. scripts/gen-catalog.mjs walks
every skills/ / /SKILL.md, reads the YAML frontmatter and writes one
entry per skill. Packages are **derived from the skills/ folders** at generation time —
a package is any top-level folder under skills/ that contains skill subfolders — so
there is no hand-maintained package or skill list.
CI runs node scripts/gen-catalog.mjs --check on every push and PR
(.github/workflows/validate.yml), so a hand-edited catalog.json cannot land.
scripts/install.sh
--target accepts codex, claude, cursor, agents, or an explicit directory path.
Without --force, an existing target folder is left untouched and the script exits non-zero.
--dry-run prints the source and destination paths it would touch and exits without writing.
Packages are **derived from the skills/ folders at install time** — never a
hand-maintained list and never read from catalog.json. A package is any top-level
folder under skills/ that contains at least one subfolder with a SKILL.md, so a
new package or skill appears automatically. --package installs every skill in
that package:
Virtual packages: generic installs every skill; meta and ref-* are reserved
on-demand names that currently resolve to the empty set. The source path for each
skill is resolved by searching the skills/ tree rooted at the script's own
directory, so the folder name alone is enough — no catalog lookup.
scripts/validate.py
- Source
- github · https://github.com/Ottili-ONE/skills
- Browse by field
- Open source
- Traction
- 1 ★
- added
- yesterday
More in Open source
Filed under
Every tag opens the full list for that filter. Counts are the live catalogue figures.