Guide · Agent skill
How to use the SuggestSkills agent skill: install and run it
Updated · written with AI from the project's own documentation and checked against our catalogue data
SuggestSkills is an agent skill that looks at your recent work and frustration records, then proposes new skills you might be missing. This guide explains what an agent skill is, how to install SuggestSkills from the ZIP on this page, how to ask for a scan in plain language, and what it needs to run. By the end you will know how to get a shortlist of evidence-backed skill ideas and how to hand an accepted idea to the separate CreateSkill skill.
Quick start
- 1Download and unzip. Download the SuggestSkills ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder.
- 2Or upload in the Claude apps. In the Claude apps, open Settings → Capabilities → Skills and upload the skill.
- 3Install Bun. Make sure the Bun runtime is available, since the collector Tools/CollectSignals.ts needs it.
- 4Check your data folders. Confirm your work sessions, ratings, skill-gap records and skills directory exist, or set the matching flags or environment variables.
- 5Ask for a scan. Ask something like "what skills should I build based on my recent work?" The default window is 45 days.
- 6Review the shortlist. Read the header and the evidence behind each proposal, and drop any that duplicate existing skills.
- 7Hand off to CreateSkill. Take an accepted proposal to CreateSkill as a separate step to build it.
What an agent skill is and what SuggestSkills does
An agent skill is a folder containing a SKILL.md file and optional supporting files, such as scripts or intent documents, that an AI agent loads when a request matches the skill's description. SuggestSkills is one such skill. Its job is to answer one question: given what you have been doing and where you have been frustrated, is there a recurring problem that deserves its own skill and does not have one yet?
Once installed, the agent can scan your recent work sessions, low-rated frustration logs, and skill-gap records. It then proposes a ranked shortlist of new skills. Each proposal must cite the evidence behind it and is checked against existing skills so it does not duplicate them.
- It is read-only: it never creates or edits a skill itself.
- Accepted proposals are handed off to a separate skill called CreateSkill, which builds or improves skills.
- Gaps hidden under topics that look covered can surface, because frustration is weighted above topic frequency.
Install the SuggestSkills skill
Download the SuggestSkills ZIP from this page and unzip it. You can install it in one of two places, depending on how you use your agent.
For a personal install that applies everywhere, unzip the folder into ~/.claude/skills/. For a single project, unzip it into that project's .claude/skills/ directory. If you use the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Personal (all projects): ~/.claude/skills/SuggestSkills
- Single project: .claude/skills/SuggestSkills inside the project folder
- Claude apps: Settings → Capabilities → Skills, then upload the skill
~/.claude/skills/SuggestSkills/Ask for a skill gap scan in plain language
You do not need to name the skill or a command. The skill triggers on requests that match its description. Phrases from its trigger list include: "should I create a skill", "what skills do I need", "suggest skills", "skill gap", "based on my recent work", "am I missing a skill", and "what should I build".
The default window is the last 45 days, with a rating cut of 4 for low-rated frustration entries. You can ask for a different window in your request, and the skill's collector accepts a days value from 1 to 3650.
- "What skills should I build based on my recent work?"
- "Skill gap: what am I missing?"
- "I keep hitting the same wall. Am I missing a skill?"
Know what the scan needs before you run it
The skill reads no config file. It takes its inputs from the Bun-run collector, Tools/CollectSignals.ts, using flags or environment variables. The root defaults to $HOME/.claude, and each store (ratings, work sessions, skill-gap records, skills directory, and an optional loops directory) can be pointed elsewhere.
If you have no recorded work history, ratings, or skill-gap data, there is little evidence to ground proposals in, so the shortlist will be thin. If the ratings store is missing, the report should say the frustration signal was unavailable for that run, rather than presenting a topic-only result as complete.
- Bun runtime, needed to execute Tools/CollectSignals.ts
- Local work-session, Errata ratings, skill-gap and skills directories (default root $HOME/.claude)
- Root override: --root <dir> or SKILLSCAN_MEMORY_ROOT
- Window override: --days <n> (default 45)
- Rating cut override: --max-rating <n> (default 4)
- Store overrides: --ratings, --work, --skill-gaps, --skills, --loops, or matching SKILLSCAN_* environment variables
- The full flag table is in Intents/Scan.md under Tool contract.
Read the report and act on it
The report header states the window, the session count, and whether the frustration store was present. Each proposal lists the evidence it rests on: sessions, frustration rows, or skill-gap detections. Read the evidence before deciding. A proposal with weak or missing supporting signals, or one that duplicates an existing capability, should be dropped.
When you accept a proposal, take it to CreateSkill as a separate step. SuggestSkills will not build it for you.
- Check the window and session count in the header.
- Confirm each proposal cites real evidence.
- Reject proposals that duplicate an existing skill.
- Send accepted proposals to CreateSkill.
Safety notes: scripts and network
SuggestSkills runs a script: Tools/CollectSignals.ts, which requires Bun. The skill does not use the network. It reads local directories you already have, and it does not edit or create files in your skills folder. Because it runs a script, review the source of Tools/CollectSignals.ts before you install, as you would with any skill that executes code on your machine.
- Runs scripts: yes (Bun)
- Uses network: no
- Read-only with respect to skills
What people use it for
- Finding recurring agent failures that have no dedicated skill yet.
- Spotting gaps hidden under topics you already think are covered, using frustration records.
- Reviewing a personal collection of skills to see which recurring problems lack one.
- Running a scan over a chosen window, such as the last 90 days, to compare what has changed.
Frequently asked questions
How do I install the SuggestSkills agent skill?
Download the SuggestSkills ZIP from this page and unzip it into ~/.claude/skills/ for a personal install, or into a project's .claude/skills/ folder for one project. In the Claude apps, you can instead upload it under Settings → Capabilities → Skills.
Is SuggestSkills free and what is its licence?
The skill is released under the MIT licence, as listed on this page. This page does not state pricing, so check the source repository for any terms beyond the licence.
What does SuggestSkills require to run?
It requires the Bun runtime to execute Tools/CollectSignals.ts, plus local work-session, Errata ratings, skill-gap and skills directories. The default root is $HOME/.claude, or you can set a path with flags or environment variables.
Does SuggestSkills create or install skills automatically?
No. It is read-only and only proposes. Accepted proposals must be taken to the separate CreateSkill skill, which builds or improves skills.
Why does SuggestSkills say the frustration signal was unavailable?
The report states this when the ratings store is missing for the run. Without it, the shortlist is based on topics and work history alone, so it is not a complete result. Point the tool at your ratings directory with --ratings or SKILLSCAN_RATINGS_FILE.
How is SuggestSkills different from CreateSkill?
SuggestSkills finds which new skills you might need and explains the evidence. CreateSkill creates, validates, or optimizes an individual skill. Use SuggestSkills to decide what to build, then CreateSkill to build it.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
SuggestSkills suits people who maintain their own skills collection, use a personal work-log setup such as LifeOS, or track repeated agent failures and want evidence for what to build next. It is a poor fit if you have no recorded work history or ratings, or if you expect it to build or install skills automatically; for building a single skill, use CreateSkill.
Download the skill