test-engineer · gupsammy — Claude Code subagent: what it does, how to install
test-engineer generates tests for critical business logic right after a feature or bug fix, skipping tests written purely to inflate coverage numbers.
no license ★ 276 updated 2026-10-29
/plugin marketplace add gupsammy/Claudest
/plugin install claude-coding@ClaudestWhat test-engineer does
The subagent runs in two modes. Advisor handles quick questions — unit versus integration, how to mock a dependency, whether existing tests are brittle — by grounding answers in the project's own conventions rather than generic testing rules. Generator mode kicks in on explicit requests like writing tests or running a gap analysis, and also fires automatically right after new code lands, scoped to the diff.
In Generator mode it first reads manifest files such as package.json, pyproject.toml, Cargo.toml or go.mod to detect the stack and test runner, then judges project maturity to decide how aggressive to be: an early-stage MVP gets tests for the two or three paths that would break the business, a mature codebase gets strategic gap-filling instead of blanket coverage. Gaps get ranked from critical (untested business logic) down to low (simple utilities nobody needs to test).
It runs the tests it writes through Bash, but only with read-only commands — git diff, git log, test runner calls — never rm, mv, git commit, or package installs like npm install. If a test fails for lack of environment variables or infrastructure, it reports the gap instead of touching the implementation. Code quality review and performance benchmarking sit outside its scope, handled by separate code-auditor and performance-auditor agents in the same repository.
What it helps with
- find untested critical paths after shipping a feature or fix
- write tests for business logic instead of chasing a coverage number
- decide between a unit test and an integration test
- judge a project's test maturity and set a phased plan
- flag brittle or redundant tests in an existing suite
Example prompts for the agent
- Write tests for the payment flow I just implemented
- Run a test gap analysis on the auth module
- Should this UI formatting function even have a test?
How to install test-engineer
1Plugin marketplace
/plugin marketplace add gupsammy/Claudest/plugin install claude-coding@Claudest2Manually: put the files in place
.claude/agents/test-engineer.mdFor all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention
Manually: put the files in place
.codex/agents/test-engineer.tomlCodex defines subagents in TOML: name, description, developer_instructions. Move the Markdown text into developer_instructions by hand. For all projects use ~/.codex/agents/
Manually: put the files in place
.cursor/agents/test-engineer.mdCursor also reads .claude/agents/
Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean
Questions about test-engineer
How to install test-engineer in Claude Code?
Run /plugin marketplace add gupsammy/Claudest, then /plugin install claude-coding@Claudest.
What does test-engineer do?
It scans changed code for untested critical business paths and writes tests for them, or gives advisor-style guidance on test strategy without generating code.
How does test-engineer differ from code-auditor?
test-engineer only handles testing — coverage gaps and test generation. Code quality review is a separate agent, code-auditor, and performance checks go to performance-auditor.
Similar subagents
All subagents for Testing and QA →test-engineer by other authors comes first.