regent-skills
Skill4 starsChanged 6 months agoApache-2.0View on GitHub
Discover and install re_gent skills. Use when the user asks what re_gent skills exist, wants to add one, or asks a question about their own history that no installed skill covers.
In this repository14 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- blameSkill
- logSkill
- showSkill
- blameSkill
- bug-blameSkill
- context-primerSkill
- file-couplingSkill
- logSkill
- rewindSkill
- showSkill
- skill-factorySkill
- style-factorySkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 37
If nothing fits, say so and offer skill-factory, which writes a new one.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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
Find the right re_gent skill and install it.
This is the skill that installs skills. It ships with rgt init so a project
that has no other skills can still find them: without it, the catalog exists but
nothing inside the agent knows to look.
1. See what is on offer
rgt skill listThe list comes from the project's re_gent server when one is configured, and
from this rgt binary otherwise. It names which, so say which you are reading —
a teammate's published skill only appears in the first case.
2. Choose by the question, not the name
Map what the user is actually trying to do:
| They are asking | Skill |
| "why is this line here", "what broke this" | bug-blame |
| "what usually changes with this file" | file-coupling |
| "what do I need to know before editing this" | context-primer |
| "which prompt wrote this line" | blame |
| "what did the agent do" | log, show |
| "what should I automate" | style-factory |
| "build me a skill that…" | skill-factory |
If nothing fits, say so and offer skill-factory, which writes a new one. Do not
install something approximate and hope.
3. Show the grant, then install
rgt skill install <name> [<name>...]The command prints each skill's allowed-tools as it writes it. Read that line
back to the user. A SKILL.md is executable instruction, not documentation:
its grant decides what it may run, and a skill fetched from a server is exactly
the case where the user should be able to say no first.
Flags worth knowing:
--agent codexwrites to.agents/skills/instead of.claude/skills/. Use it when the user works in Codex, or the file lands where nothing loads it.--forcereplaces a skill the user has edited. Never pass it unprompted; the default refusal exists because their edit outranks the shipped copy.--server <url>installs from a specific registry.
4. Say what has to happen next
Agents load skills at startup, so a freshly installed skill does nothing until the session restarts. Tell the user plainly. Do not claim the skill is usable in the current conversation, because it is not.
Report
What you installed, where each file went, the tool grant for each, and the restart. If you installed nothing, say why — an unclear request is a reason to ask, not to guess.
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.

