tstack
A collection of agent skills for Claude Code and Codex that automate pull request workflows such as filing, reviewing, babysitting, and merging.
At a Glance
About tstack
tstack is an open-source collection of agent skills published by t3dotgg (Theo Browne holds the MIT license copyright). The skills are the author's daily-use workflows, cleaned up for general use in Claude Code and Codex. Most center on taking code changes through pull requests, review, and merge.
What It Is
tstack is a set of skill folders you copy into your Claude Code or Codex skills directory. Each skill is triggered by a short phrase, such as "full send", "council", or "babysit". Some skills use T3 Code features when available and fall back to native tools otherwise.
Skills and how they chain
The skills cover the pull request lifecycle. full-send takes a change to a merged PR unattended: it proves the change, reviews it, files it, babysits it, and merges it. full-stop does the same but stops when the PR is ready for you to merge. council gets review from three models from different providers and loops until they agree, while sol-loop loops a review with GPT Sol until both agents agree. file-pr writes a short PR that leads with the problem. takeover finishes someone else's PR, branch, or agent thread while keeping their credit. babysit-pr watches CI and review bots and fixes real findings. modernize rebases an old PR and decides whether it should still exist. leaving-pr-comments formats comments an agent posts on your behalf.
Skills depend on each other: full-send uses sol-loop, file-pr, and babysit-pr; full-stop uses full-send; takeover uses modernize, file-pr, babysit-pr, and leaving-pr-comments. Install the dependencies too.
Setup and requirements
Installation is cloning the repository and copying skills/* into ~/.claude/skills/ or ~/.codex/skills/. An authenticated GitHub CLI (gh) is required, and python3 is needed for the pr-wait script in babysit-pr. council and sol-loop need subagents on other providers' models, which T3 Code handles with delegate_task; otherwise a harness that can run those models is needed.
Customization
The skills are meant as starting points to edit. Personal settings, such as which models sit on your council, can go in CLAUDE.md or AGENTS.md so skills can be updated without losing them.
Community Discussions
Be the first to start a conversation about tstack
Share your experience with tstack, ask questions, or help others learn from your insights.
Pricing
Open Source (MIT)
Agent skills for Claude Code and Codex, released under the MIT License and free to use, copy, modify and distribute.
- MIT License
- Works with Claude Code and Codex
- Skills: full-send, full-stop, council, sol-loop, file-pr, takeover, babysit-pr, modernize, leaving-pr-comments
- Requires authenticated GitHub CLI (gh) and python3 for babysit-pr
Capabilities
Key Features
- full-send skill takes a change to a merged PR unattended
- full-stop skill stops when the PR is ready to merge
- council skill gets multi-provider model review until agreement
- sol-loop skill reviews changes with GPT Sol in a loop
- file-pr skill writes short problem-led PRs
- takeover skill finishes someone else's PR while keeping credit
- babysit-pr skill watches CI and review bots and fixes real findings
- modernize skill rebases old PRs and decides if they should still exist
- leaving-pr-comments skill formats agent-posted comments
- Editable skills with personal settings kept in CLAUDE.md or AGENTS.md
