Content
81%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.
The body is a well-sequenced, validation-heavy orchestration workflow with concrete commands and explicit feedback loops, written tersely without padding. Its main structural weakness is that everything lives in one ~100-line file with no reference material split out, and a few steps delegate to external skills without invocation syntax.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense, imperative, and assumes competence — no tutorial padding or explanations of known concepts ("Use the installed Superpowers skills for their respective stages"). Minor tightening is possible: the "Explicit invocation only" section duplicates frontmatter policy, and dense policy sentences like "Automatic watching is useful feedback, but a queued job is not a pass" could be trimmed, matching anchor 4 rather than anchor 5. | 4 / 5 |
Actionability | Concrete, executable detail is present: the full `roborev review --from-skill --type goal --agent <reviewer> --wait --spec <spec-path>` command, a complete bash example with concrete paths, the `POST /api/goal-review` endpoint, and named external skills (writing-plans, verification-before-completion). It stops short of anchor 5 because many steps ("Use Superpowers brainstorming", "claim its Kata issue") give direction rather than copy-paste commands, but it clearly exceeds anchor 3's pseudocode level. | 4 / 5 |
Workflow Clarity | The multi-step process is explicitly sequenced (brainstorm → spec → goal review → resolve findings → plan → linked review → execute → complete) with validation checkpoints and feedback loops throughout: "Request a new goal review after edits", "observe the intended test fail", "verify each finding, fix valid in-scope defects with regression tests", "Await required CI and current-head reviews before claiming them green". This matches anchor 5's explicit validation and error-recovery loops. | 5 / 5 |
Progressive Disclosure | No bundle files exist, and the ~100-line body is organized into well-ordered sections (Usage, invocation policy, intent-to-work, execute/review, completion, sandbox access) with command details inline where needed. It matches anchor 4 (good structure, most content appropriately placed, minor organization gaps); it falls short of anchor 5 because the skill exceeds 50 lines and some self-contained material (sandbox troubleshooting, the Kata integration endpoint contract) could plausibly live in a reference file. | 4 / 5 |
Total | 17 / 20 Passed |