Content
86%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A well-structured, actionable skill body that keeps SKILL.md as an overview and pushes run-result and runner detail into real, one-level-deep reference files. Conciseness and inline workflow checkpoints are the only minor gaps.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly lean and high-signal, assuming Claude's competence (no padding about what a browser or CLI is), with a few explanatory passages — e.g. the billing and --json-stability prose — that could be tightened slightly. Not a 5 because some prose in Auth/Output/Safety restates reasoning a model could infer. | 4 / 5 |
Actionability | Provides a full command table of exact `qawolf <command>` invocations plus specific flags where they matter (--json, --env, --environment-id, --runner, QAWOLF_RUNNER_ID); deferring remaining flags to `--help` is explicitly justified. Common cases are copy-paste ready. | 5 / 5 |
Workflow Clarity | Clear sequences for auth/environment resolution, the runner see-and-act loop, and git publishing, with safety checkpoints ('Never blind-retry a write on timeout', 'a successful write response is confirmation'). Not a 5 because the full runner ordering is deferred to references/runner.md rather than given inline with explicit validate-fix-retry loops. | 4 / 5 |
Progressive Disclosure | Clear overview with well-signaled one-level-deep references to references/run-results.md and references/runner.md (both present in ./references/); detail is appropriately split out of SKILL.md and the commands table is kept inline as a quick reference. | 5 / 5 |
Total | 18 / 20 Passed |