Content
85%Weight 40%Scale 1-3Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is a well-structured, token-efficient overview with clear sequencing and recovery handling, and it appropriately offloads implementation to the bundled detect script; its only gap is that the body itself lacks concrete executable commands or invocation examples.
Suggestions
Add a short "Running detect" snippet showing the actual `gws`/script invocation and its key flags so the body is actionable without reading detect.ts.
Include one minimal example of a watch query and the labeled-match output handed to the drafting turn.
Show the `gws gmail +reply` approval-gate command in full rather than only naming it.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes competence — it covers only skill-specific behavior (watermarks, dedup, per-watch independence, first-run baselines) without explaining what Gmail or polling is, and every section earns its place. | 3 / 3 |
Actionability | It references the bundled `detect` script and mentions `gws gmail +reply`, but the body itself gives no executable commands or invocation syntax — concrete guidance is incomplete because the real logic lives in detect.ts. | 2 / 3 |
Workflow Clarity | A clear sequenced workflow is laid out (the numbered "What it does" steps and "How a match becomes a reply") with explicit failure/recovery checkpoints (needs_auth, per-watch transient errors, no silent loss, first-run baselining). | 3 / 3 |
Progressive Disclosure | The body is a concise overview that correctly defers the heavy logic to the real bundled file scripts/detect.ts, with clearly signaled one-level-deep references and no nested indirection. | 3 / 3 |
Total | 11 / 12 Passed |