lions
Skill2 starsChanged 45 days agoMITView on GitHub
Use when the user invokes /lions or explicitly asks every other model and account to do the same task separately, each from its own view (a code review, an analysis, a second opinion). For splitting a job into parts use voltron; for a deliberation use cogitors.
In this repository5 other files it gives its agents.
- antigravitySkill
- claudeSkill
- codexSkill
- cogitorsSkill
- voltronSkill
The file
Files
Lions
The lions set out separately: every lion does the same whole task from its own view, and the Elder compares what they bring back and gives the user one answer. Read lions first, for who the lions are, choosing models, the plan format, running and reading results; in a repository checkout it is ../cogitors/references/lions.md.
Send the lions
- Write the task for a
sameplan: the user's task in their words, the scope, and the output you need back (for a review: findings withfile:line, severity and evidence). - Choose lions and models. By default every available lion goes. Pick each one's model and effort from
cogitor.py models; prefer different model families, since the point is independent views. - Announce the lions, models and access in one short block, then start right away unless the user asked to see the plan first.
- Run the
sameplan with--access review, oreditwhen each lion should produce its own version of a change to compare.
Answer
Give the user one answer in their language. Account for every finding: merged with its duplicates and counted by how many lions raised it (3/4), verified by you, labeled as unconfirmed, or dropped with the reason. Show raw reports only when the user asks. For edit, compare the lions' diffs and apply the best one, or combine them. Then clean the run.
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.