Claude Code
AnthropicAnthropic's terminal agent: reads the repo, edits files, runs the tests it just broke.
Turn an incident list into a decision about what to fix next.
Teams with an incident log and more possible fixes than capacity
A single incident — postmortems handle that
Reads incidents and error-budget consumption and ranks the next piece of reliability work by frequency times cost. The output is a short list with a reason per item, which is the form a planning meeting can act on. It does not estimate effort; that is the team’s input.
Clone the repository, then copy the skill folder into your agent's skills directory — ~/.claude/skills/ for Claude Code. Skills are plain files: no build step, no daemon.
npx skills add error-budget-reviewRead the SKILL.md before you install. A good skill states when it applies, what it must not do, and how to verify the result. A skill that only lists adjectives will change the agent's tone without improving its output — skip those.
Anthropic's terminal agent: reads the repo, edits files, runs the tests it just broke.
Code assistant that answers with repository context across many repos.
Self-hosted code completion with an open-source server.
Gives the model a scratchpad for structured, revisable reasoning: numbered thoughts that can branch, revise, and conclude. It does not make a weak model strong, but …
Serves version-specific documentation for thousands of libraries, so an agent stops writing code against an API that changed two releases ago. The practical fix for …