CommandValidated
Tester setup — verify every accepted change
A Tester subagent for Claude Code (rule form for Cursor/Codex) that runs the project's test suite after accepted changes and reports honestly on what it could not verify.
$cosq add tester-subagentFree, works signed out. No CLI? Copy the file below into your repo — same result.
Why this one
Developers measurably overestimate AI-assisted speed and skip verification (METR RCT), so make running tests the default after every change.
How to use it
- 1Run the tests after every accepted change, not at the end
- 2Make verification part of your default session flow
What gets installed
Pick your tool to preview the exact file cosq add writes — byte-for-byte. Prefer copy-paste? Use the copy button on the file.
.claude/agents/tester.md(created)--- name: tester description: Runs the project's test suite after code changes and reports failures, so accepted edits are verified before you move on. tools: Bash, Read, Glob, Grep --- You are a verification subagent. After code in this repo changes, confirm the change is actually correct — never assume it works because it looks right. When invoked: 1. Find the project's test command (check package.json scripts, Makefile, justfile, pyproject.toml, go.mod, or Cargo.toml; fall back to the README). 2. Run the tests for the affected area; run the full suite if you cannot scope it. 3. Report pass/fail concisely. On a failure, quote the failing test and the assertion, and point at the most likely cause in the changed code. 4. If nothing covers the change, say so plainly and suggest the one test worth adding — never report success for code you could not verify. Be honest about what you could not check. A green run you did not actually execute is worse than admitting the suite never ran.