/find-skills
Find, compare, audit, and safely install reusable AI agent skills from OpenAgentSkill. Use when a user asks for a skill, plugin, reusable agent workflow, or the best tool for a task.
$ npx -y skills add Leon-Drq/openagentskill --skill find-skills --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
- Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
- You can call itInvoke it directly when you want it.
- Slash command
/find-skills
Context preview
The summary Claude sees to decide when to auto-load this skill.
Find, compare, audit, and safely install reusable AI agent skills from OpenAgentSkill. Use when a user asks for a skill, plugin, reusable agent workflow, or the best tool for a task.
SKILL.md
find-skills.SKILL.mdname: find-skills
description: Find, compare, audit, and safely install reusable AI agent skills from OpenAgentSkill. Use when a user asks for a skill, plugin, reusable agent workflow, or the best tool for a task.
license: MIT
compatibility: Requires internet access to www.openagentskill.com. Installation commands require Node.js 20+.
metadata:
author: OpenAgentSkill
homepage: https://www.openagentskill.com/resolve
Find Skills with OpenAgentSkill
Turn the user's task into an evidence-backed shortlist. Do not recommend a skill only because its repository is popular.
Workflow
1. Resolve the task with the public endpoint:
`GET https://www.openagentskill.com/api/agent/resolve?task=<encoded-task>&agent=<agent-name>`
2. Read `selected`, `recommendation_lanes`, `policy_decision`, and at least three entries from `alternatives` when they exist. 3. Before installing, open the selected Skill's `urls.audit` and inspect its repository source, license, permissions, setup requirements, and freshness. 4. Use `install_plan.command` only when the policy allows it. If human review is required, present the risks and ask for approval. Never install a blocked Skill. 5. Run one narrow verification task in a sandbox or low-risk workspace. 6. Report the result to the returned feedback endpoint with its unique `event_id`. Report success only when installation and the verification task both succeeded.
CLI
The official release can perform the same workflow:
`npx --yes https://github.com/Leon-Drq/openagentskill/releases/download/cli-v0.3.0/openagentskill-0.3.0.tgz find "<task>"`
Use `resolve` for the complete decision object and `add <slug> --dry-run` before a reviewed install. Anonymous telemetry can be disabled with `--no-telemetry` or `OPENAGENTSKILL_TELEMETRY=0`.
Decision rules
- Prefer task fit, safety, verified outcomes, and maintenance over raw stars.
- Treat repository stars as a project-level popularity signal, not proof that
every nested Skill has equivalent adoption.
- If no candidate fits, say so and use the agent's native tools instead of
forcing an unrelated recommendation.
Read more
name: find-skills description: Find, compare, audit, and safely install reusable AI agent skills from OpenAgentSkill. Use when a user asks for a skill, plugin, reusable agent workflow, or the best tool for a task. license: MIT compatibility: Requires internet access to www.openagentskill.com. Installation commands require Node.js 20+. metadata: author: OpenAgentSkill homepage: https://www.openagentskill.com/resolve
Find Skills with OpenAgentSkill
Turn the user's task into an evidence-backed shortlist. Do not recommend a skill only because its repository is popular.
Workflow
1. Resolve the task with the public endpoint:
`GET https://www.openagentskill.com/api/agent/resolve?task=<encoded-task>&agent=<agent-name>`
2. Read `selected`, `recommendation_lanes`, `policy_decision`, and at least three entries from `alternatives` when they exist. 3. Before installing, open the selected Skill's `urls.audit` and inspect its repository source, license, permissions, setup requirements, and freshness. 4. Use `install_plan.command` only when the policy allows it. If human review is required, present the risks and ask for approval. Never install a blocked Skill. 5. Run one narrow verification task in a sandbox or low-risk workspace. 6. Report the result to the returned feedback endpoint with its unique `event_id`. Report success only when installation and the verification task both succeeded.
CLI
The official release can perform the same workflow:
`npx --yes https://github.com/Leon-Drq/openagentskill/releases/download/cli-v0.3.0/openagentskill-0.3.0.tgz find "<task>"`
Use `resolve` for the complete decision object and `add <slug> --dry-run` before a reviewed install. Anonymous telemetry can be disabled with `--no-telemetry` or `OPENAGENTSKILL_TELEMETRY=0`.
Decision rules
- Prefer task fit, safety, verified outcomes, and maintenance over raw stars.
- Treat repository stars as a project-level popularity signal, not proof that
every nested Skill has equivalent adoption.
- If no candidate fits, say so and use the agent's native tools instead of
forcing an unrelated recommendation.
The skill layer for AI agents. Find, compare, audit, and install the right reusable Agent Skill before an agent acts.
Repo: Leon-Drq/openagentskill