morluto/internalcot
Make agents show their full chain of thought.
internalcot makes agents show their full chain of thought. Turn it on once and the agent must externalize its problem decomposition, intermediate derivation, alternatives, evidence, uncertainty and checks in the normal tool transcript for the whole conversation — producing a persistent, readable chain of thought you can inspect as the agent works, instead of a summary that hides the reasoning. It is an opt-in reasoning workspace: the README's example shows the agent restating a proof problem, the derivation, and an explicit next check, with the reasoning written into visible working notes rather than kept private.
Install
npx internalcot@latest setupThe README's command is 'npx internalcot@latest setup'; it says the setup installs the CLI and the skill together, detects Codex and Claude Code, shows the exact global command and skill paths, and asks before changing anything (a restart may be needed for the skill to appear). An unattended Codex form (--codex --yes, with a --dry-run variant) and a skill-only form ('npx skills add morluto/internalcot') are also documented. The npm package internalcot 0.2.3 was re-verified live on 2026-10-02 with its repository field back-linking to morluto/internalcot (MIT).
Compatibility
A shell CLI plus an agent skill; the README targets Codex and Claude Code hosts. When the persistent CLI is unavailable the skill can run the current package through 'npx --yes internalcot@latest skill --npx'. Setting up changes global command and skill paths, and the wizard asks before applying them.
Details
- Repo: morluto/internalcot
- Category: Agent Capabilities
- Stars: 10
- Version: npm internalcot 0.2.3 (registry re-verified 2026-10-02); MIT
- Last push: 2026-08-11
- First seen: 2026-08-11
FAQ
- How do I install internalcot?
- npx internalcot@latest setup — it installs the CLI and the skill together, detects Codex and Claude Code, shows the exact paths and asks before changing anything.
- What does it change about the agent's behaviour?
- It adds an opt-in reasoning workspace: the agent must externalize decomposition, derivation, alternatives, evidence, uncertainty and checks in the visible transcript.
- Can I install just the skill?
- Yes — the README documents 'npx skills add morluto/internalcot'; if the persistent CLI is unavailable the skill runs the current package through npx.
Alternatives
gongyijie85/mattpocock-skills-dsh · hackerFish/awesome-dsh-skills · JazzuLu/find-dsh-plugins