skill-issue
/skill-issue
Find which installed Agent Skills fit the current task by asking the local skill-issue router, and handle its suggestions safely. Use at the start of any non-trivial task when no <skill-issue> context block is already present, and whenever the user asks why a skill did or did not trigger, wants to switch skill-issue modes, or wants to install a skill the router suggested.
- Author
- sahilbachu ↗
- Repository
- skill-issue ↗
- Licence
- Apache-2.0
- Last repository change
- 26.09.2026
Assessment and security
- Files in this package
- 1
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 26.09.2026, without references to retired models or APIs.
Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.
What it does and when to use it
Find which installed Agent Skills fit the current task by asking the local skill-issue router, and handle its suggestions safely. Use at the start of any non-trivial task when no <skill-issue> context block is already present, and whenever the user asks why a skill did or did not trigger, wants to switch skill-issue modes, or wants to install a skill the router suggested.
Original author description
Find which installed Agent Skills fit the current task by asking the local skill-issue router, and handle its suggestions safely. Use at the start of any non-trivial task when no <skill-issue> context block is already present, and whenever the user asks why a skill did or did not trigger, wants to switch skill-issue modes, or wants to install a skill the router suggested.
How it works
- When a <skill-issue> block is already in context
- How to consult the router
- Skills that are not installed (approved mode)
- Other commands the user may ask for
SKILL.md preview
---
name: skill-issue
description: Find which installed Agent Skills fit the current task by asking the local skill-issue router, and handle its suggestions safely. Use at the start of any non-trivial task when no <skill-issue> context block is already present, and whenever the user asks why a skill did or did not trigger, wants to switch skill-issue modes, or wants to install a skill the router suggested.
---
# skill-issue: consult the skill router
skill-issue is a local router. Given a request, it returns the few skills worth loading, or none.
Returning none is normal and common. It runs on this machine; nothing is sent anywhere.
## When a `<skill-issue>` block is already in context
The Claude Code plugin already routed this prompt. Do not call the router again. If a listed
skill fits, load it before starting. Ignore listed skills that do not fit.
## How to consult the router
Prefer the MCP tool if it is available:
- `find_skills(request, cwd)` returns `{"skills": [...]}`. Each installed skill has a
`skill_md` path. Read that file and follow it.
- `explain_route(request, cwd)` shows every candidate with its probability and retrieval rank.
Use it when the user asks why something was or was not picked.
Otherwise use the CLI:
```bash
skill-issue route "<the user's request, verbatim>" --json
```
Read `selected`. For each selected skill with `"installed": true`, read `<path>/SKILL.md` and
follow it. If `selected` is empty, continue without a skill. Do not keep retrying with reworded
requests to force a match.
## Skills that are not installed (approved mode)
In approved mode the router can suggest a vetted public skill marked `"installed": false`.
1. Never install it on your own.
2. Show the user its name, source repo and commit, trust tier, license, sha256, and the list of
scripts it contains. Say that installing copies those files into their skills folder.
3. Only if the user explicitly says yes, run:
`skill-issue install <name> --confirm <first 12 characters of the sha256>`
4. The command re-verifies the hash and refuses on any mismatch. Report what it printed.
## Other commands the user may ask for
| Request | Command |
|---|---|
| Switch modes | `skill-issue mode installed` or `skill-issue mode approved` |
| Check the setup | `skill-issue doctor` |
| See what the router can see | `skill-issue index` |
| Scan a skill folder for risky content | `skill-issue scan <path>` |
| Browse the approved index | `skill-issue approved <search words>` |
`skill-issue scan` only reads files. Never run a third-party skill's scripts to "test" it.
Versions
| Repository | Score | Licence |
|---|---|---|
| This page · sahilbachu/skill-issue | 100 | Apache-2.0 |
| eliasstravik/skills | 90 | MIT |
Database snapshot: 13.11.2026 · package version 49915b00a65d (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 90skill-library-router
Find and safely apply a specialized Agent Skill from an indexed GitHub repository catalog when installed skills do not cover the task, or when the user asks to search, refresh, or install cataloged skills. - 100skill-recall
Find local skills for the current task and log gaps. Use when checking which local skill matches a task or whether a suitable skill exists. - 100open-skill-router
Picks which installed skills to use for a task and in what order, from the user's IT role and the project's state (spec-kit, BMad, superpowers, codegraph), then starts the first skill. Use when starting a new project or multi-step work (specify, plan, build, test, review) with… - 90skillme
Find, install, and manage Skill Me skills from inside a conversation through the Skill Me connector. Use when the user asks for a Skill Me skill, asks what skills they have, wants skills suggested for a task, or wants to install or remove a skill or pack. - 100skill-router
This skill should be used when the user has a vague or cross-domain request and asks "which skill should I use", "is there a skill for this", "find the right skill", or "route this to the best skill" — it matches an intent against the whole skills library and recommends the best-fit skill(s). - 100skill-router
Use when the user is unsure which skill to use or where to start. Interviews the user with targeted questions and recommends the best skill(s) from the installed library for their goal.