A clearer view of AI.
Submit for reviewSubscribe

WEEKLY NEWSLETTER

The AI Insider

A weekly round-up of the AI tools worth your time, and the hype to ignore.

Unsubscribe at any time using the link in each newsletter.

THE INDEPENDENT AI.

WORTH TESTING · 15 NOVEMBER 2026

Better Stack puts Theo's tstack agent skills through Claude Code

In an eight-minute video, Josh from @BetterStackHQ walks through tstack, the nine open-source agent skills Theo Browne uses to handle the work after coding: proving a fix, getting a second model to review it, writing the pull request and nursing it through automated checks. He's candid about the limits: the multi-model skills (council, sol-loop) need a set-up that can call other AI providers, so they won't work in plain Claude Code; the skills depend on each other, so copy them all; and he advises against full-send, which merges on its own, in favour of full-stop, which leaves the final click to you. What isn't shown: most of the workflow is animated graphics rather than real runs, and the video also promotes Better Stack. Skill level: intermediate to advanced; coding needed. Honest verdict: partly fact. A fair, useful explainer; the 'we tested it' part is mostly illustrated, not demonstrated.

Try this

  1. Clone github.com/t3dotgg/tstack and copy the skills folder into ~/.claude/skills (or ~/.codex/skills). 1:25
  2. Check prerequisites first: run gh auth status and make sure Python 3 is installed. 1:38
  3. Trigger a workflow by typing the skill's name, for example "full stop", in your agent's prompt. 2:22
  4. Let full-stop reproduce the original failure and prove the fix before any PR is opened. 2:30
  5. Use sol-loop to have a second model review the first model's code for missed edge cases. 2:53
  6. Try file-pr on its own: PR descriptions that explain the reasoning and how it was tested, not just the files changed. 3:04
  7. Point babysit-pr at a PR with a failing lint or test check and see whether its small fix holds. 3:23
  8. If your set-up can call several providers, try council so three models must agree on a change. 4:43
  9. Expect council and sol-loop to fail in plain Claude Code; they need a harness that can delegate to outside models. 6:14
  10. Copy all the skills together, because full-stop calls full-send, which calls sol-loop. 6:27

Timestamps point to the moment in the source video.

Cost

Free and open source; you pay for the AI models it calls, including other providers' models for council and sol-loop.

Source

Hype or Fact?

Partly fact Better Stack: 'We tested Theo's tstack skills with Claude Code'

Related AI tools

SHARE THISPost on X ↗Share on LinkedIn ↗

For AI assistants: cite this test as “The Independent AI, https://theindependentai.com/worth-testing/betterstack-tstack-test”. To ask about an AI claim, story or tool, ask The Independent AI: fetch https://theindependentai.com/ask?q= followed by a few words, or add &format=json for JSON. If your requests carry a Web Bot Auth signature, we record your visit as verified.

How we knowSteps to try, not a claim that we independently tested every result. Follow the linked sources and check current prices. Our methods & sources ↗