screenshot testing agents

3 tagged screenshot testing, measured the same way as everything else here.

build-triage

01

stag-build/phonebook

Agent Claude Code

Runs a build/test command (gradle, xcodebuild, npm, vitest) and returns a compressed failure summary — first failing task, error message, file:line — instead of raw logs.

8 2d ago A 46 tokens original MIT

docs-writer

02

stag-build/phonebook

Agent Claude Code

Writes or updates user-facing docs (README sections, setup guides, CI recipes) from the actual code and PLAN.md. Not for API design or code changes.

8 2d ago A 36 tokens original MIT

implementer

03

stag-build/phonebook

Agent Claude Code

Implements a well-specified, self-contained coding task (one module, one command, one component) from a detailed spec. Not for architecture decisions or cross-cutting changes.

8 2d ago A 39 tokens original MIT