skill-inventory
Skill1 starsChanged 3 days agoMITView on GitHub
Lists every Agent Skill installed for the current user and project across Claude Code, Codex, Gemini CLI and Antigravity, and reports duplicate names, skills that fail to load, descriptions that compete for the same request, and the context cost of the skill listing. Use when asked which skills are installed or available, when the wrong skill fires or two skills conflict, when a skill seems to be shadowed or missing, or when cleaning up, pruning or deduplicating skills.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- Mathews-Tom/armorytheir own version, first seen Nov 1, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- Mathews-Tom/armory★ 340same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository6 other files it gives its agents.
- skill-auditSkill
- skill-distillSkill
- skill-lintSkill
- skill-portSkill
- skill-retroSkill
- skill-tuneSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 46
Always-loaded listing cost: Every installed skill's name and description is placed in the context of every session.
- Distinctive: line 50
Same name in more than one place: Only one copy wins in a given harness, and which one depends on that harness's precedence rules.
- Distinctive: line 74
Before deleting or moving anything:
- Distinctive: line 77
Do not edit skills inside a plugin cache, a synced folder or a built-in directory.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Skill inventory
Answers "what skills are actually in play here, and are any of them getting in each other's way?" from the file system, not from memory of what is installed.
Run it
Paths below are relative to this skill's directory.
python3 scripts/inventory.py [--project <dir>] [--root <dir> ...] [--no-user] [--threshold 0.4] [--json]--projectis the project to inspect; it defaults to the current directory.--rootadds a directory the script does not know about. Repeatable.--no-userlimits the scan to the project.--thresholdsets how similar two descriptions must be (0 to 1) to be reported as competing. Lower it to see more candidate pairs.- The script only reads. It never changes or removes a skill.
The last section of the output, "Searched", shows each location and whether it
exists. If the user's harness keeps skills somewhere not listed there, find
that directory and pass it with --root.
Read the output
- The listing, grouped by location. A trailing
!marks a skill with a load problem. - Always-loaded listing cost. Every installed skill's name and description is placed in the context of every session. This is the estimated total. It is the price of having skills installed, paid whether or not they are used.
- Same name in more than one place. Only one copy wins in a given harness, and which one depends on that harness's precedence rules. In Gemini CLI, for example, workspace skills override user skills. Do not guess for other harnesses; check their documentation or ask the harness to list its skills.
- Skills that break the specification. Missing or invalid frontmatter, or a name that does not match its directory. Strict harnesses skip these, so a skill can be installed and still invisible.
- Descriptions that may compete. Pairs with heavy keyword overlap. This is a lexical measure; treat each pair as a question, not a verdict.
Act on it
Report findings first. Change nothing until the user agrees.
| Finding | Recommendation |
| Duplicate name, identical content | Keep one copy at the widest scope that needs it; remove the others. |
| Duplicate name, different content | Show the differences. The user picks one or renames one. |
| Broken skill | Fix the frontmatter. If the skill-lint skill is installed, use it for the exact errors. |
| Competing pair, same job | Merge them, or retire the weaker one. |
| Competing pair, different jobs | Sharpen both descriptions so each says what it is not for. If the skill-tune skill is installed, use it. |
| Large listing cost | Suggest removing skills that are never used, moving project-only skills out of the user scope, and shortening long descriptions. |
Before deleting or moving anything:
- Confirm with the user, naming the exact paths.
- Do not edit skills inside a plugin cache, a synced folder or a built-in directory. The owning tool overwrites them. Disable or uninstall through that tool.
- A symlinked skill is shared with wherever the link points. Removing the link is safe; editing through it changes the original.
Limits
- The script knows the locations listed in its "Searched" output. Enterprise
or managed locations and unusual installs need
--root. - Finding a skill on disk does not prove the harness has it enabled; a harness can disable skills through its own settings.
- Token figures are estimates (about four characters per token).
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.