Mente is a unified AI agent for coding, automation, gateway workflows, and long-term memory.
This repository also configures its own agents. See what Mente tells them →
Skill Claude CodeCodex
Use when the user asks Mente to change API keys, OAuth/provider auth, config.yaml or .env settings, gateway/platform credentials, or restart/reload services after configuration changes.
Skill Claude CodeCodex
Plan mode: write markdown plan to .mente/plans/, no exec.
Skill Claude CodeCodex
Pre-commit review: security scan, quality gates, auto-fix.
Skill Claude CodeCodex
Execute plans via delegatetask subagents (2-stage review).
Skill Claude CodeCodex
4-phase root cause debugging: understand bugs before fixing.
Skill Claude CodeCodex
TDD: enforce RED-GREEN-REFACTOR, tests before code.
Skill Claude CodeCodex
Mente Superpower: create an isolated git worktree before substantial project work.
Skill Claude CodeCodex
Mente Superpower: verify before claiming a task is done.
Skill Claude CodeCodex
Write implementation plans: bite-sized tasks, paths, code.
Skill Claude CodeCodex
Yuanbao (元宝) groups: @mention users, query info/members.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: