Runs the configured test suite and records the result for downstream use by /dev-pr. Useful for mid-task verification or debugging a failing test in isolation.
How this skill is triggered — by the user, by Claude, or both
Slash command
/claude-code-dev-hermit:dev-testThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Run the project's configured test command and record the result.
Run the project's configured test command and record the result.
Optional --cwd <path>. When set, the test command runs from <path> and last-test.json records <path>'s HEAD SHA. Use this for nested-repo workflows (see CLAUDE-APPEND §Implementation Flow). <path> must be a git working tree.
Run the following Bash command. Use timeout: 600000 (10-min ceiling). Append --cwd "<path>" when the operator passed --cwd.
bun "${CLAUDE_PLUGIN_ROOT}/scripts/record-test-result.ts" run
If exit 0: report tests: pass.
If exit 1 and stderr is "commands.test not configured": tell the operator to set commands.test in .claude-code-hermit/config.json (or re-run /claude-code-dev-hermit:hatch).
If exit non-zero for any other reason: report tests: FAIL (exit N) with the last 10 lines of output.
/dev-pr invokes the same machinery on cache miss — running /dev-test first is optional but warms the cache (a second /dev-pr call at the same HEAD skips the test run entirely).
For suites longer than 10 min (Bash ceiling): run the test command directly in a terminal, then record the result manually:
bun <CLAUDE_PLUGIN_ROOT>/scripts/record-test-result.ts write <exit_code> <duration_ms>
Then re-run /dev-pr.
npx claudepluginhub p/gtapps-claude-code-dev-hermit-plugins-claude-code-dev-hermitShows test status, coverage, flaky tests, and history from cached results without re-executing. Supports pytest, Vitest, Jest, Playwright, Go, Cargo.
Scouts test coverage gaps, creates test files, continues incomplete suites, tracks persistent coverage using project test config and git analysis.