LeoLin990405/grimoire-skill / skills/grimoire-manage
grimoire-manage
UnverifiedCompanion to Grimoire — the OPT-IN layer that takes a produced skill pack, asks the user which global agents to configure, installs it across them, and then manages it (status / list / sync / uninstall / preflight gate). Use after grimoire.sh has produced & you have reviewed a skill pack and the user wants it usable in their other agents (Claude Code / Codex / Trae / opencode / copilot / qwen / deepseek / stepfun), or wants to re-sync after a Trae/stepfun/deepseek app update, check cross-agent parity, remove a skill everywhere, or add a confirm-before-running gate. This deliberately crosses Grimoire's default "scripts never install skills" boundary by explicit user request and records it in the pack manifest. Triggers: 装 skill 到其它 agent / 把技能打通到所有 agent / 跨 agent 安装技能 / skill 没出现在某个 agent / 重新同步技能 / 卸载某个技能 / skill 确认门 / install skill across agents / cross-agent skill / sync skills after update / skill preflight gate / grimoire-manage.
Grimoire-Manage · 跨 agent 安装与管理
Grimoire produces a candidate skill pack and never installs it (red line).
grimoire-manage is the explicit, user-driven layer that installs and manages
it across agents. Two scripts (repo scripts/, mirrored here under scripts/):
skill-install.sh— install a reviewed pack into chosen agentsskill-manage.sh— status / list / sync / uninstall / gate
When to use
| Need | Command |
|---|---|
| Scan this machine: how many skill-capable agents exist | skill-manage.sh scan |
| Check the host has a source skill's config/tools (MinerU server/token, ffmpeg/yt-dlp, Kedou) | skill-manage.sh doctor [--skill <name>] |
| See agents + which exist (drive the user choice) | skill-install.sh --pack <dir> --list |
| Preview install (no changes) | skill-install.sh --pack <dir> --agents <csv> --dry-run |
| Install into chosen agents | skill-install.sh --pack <dir> --agents <csv|all> |
| Cross-agent parity of a skill | skill-manage.sh status --name <skill> |
| Per-agent counts + agent-only skills | skill-manage.sh list |
| Re-sync after a Trae/stepfun/deepseek update | skill-manage.sh sync --pack <dir> |
| Remove a skill everywhere | skill-manage.sh uninstall --name <skill> |
| Add the confirm-before-running gate | skill-manage.sh gate |
The flow (produce → choose → install → manage)
grimoire.sh … --only skills --install(or run grimoire normally) → review/promote candidates intoskills/skills/.- Inventory:
skill-install.sh --pack <skills-dir> --list. - Ask the user which agents to configure — in Claude Code use
AskUserQuestion; otherwise ask in chat. Choose onlypresent = yesrows. Never pick agents for the user. - Dry-run, confirm, install:
--dry-runfirst, then the same command without it. Tell the user to restart each affected agent. - Manage later via
skill-manage.sh(status/sync/uninstall/gate).
Boundary
- Installing crosses the default candidate-only red line by explicit user
request;
skill-install.shrecords this in the pack manifest'sred_line_note/installed_to. - Per-pack only: an agent's own non-shared skills (e.g. Codex
ppt-master, Traeweb-dev) are never enumerated, moved, or deleted; drifted same-named dirs are backed up, never clobbered. - The Claude
settings.jsonpreflight hook is printed for manual install (the agent cannot self-edit startup config);gateinjects only the portable instruction-file paragraph and never touchessettings*.json. - Verified agent topology and rules live in
scripts/lib/agent-targets.sh.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 470k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |