Content
88%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.
An exemplary operational skill: concrete commands throughout, a hard preview-then-confirm gate before a destructive multi-repo operation, per-repo error handling, and no padding. The only refinements available are trimming a few rationale asides and optionally moving the runtime-specific transcript lookup out of the main flow.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense, procedural, and assumes competence — no explaining of git or what session tracking is, with every section carrying operational rules or commands. The short 'Why this skill exists' section and a few explanatory asides in Steps 2-3 (e.g. the rationale for keeping misconfigured candidates) are trimmable, matching anchor 4's 'minor instances of over-explanation'. | 4 / 5 |
Actionability | Guidance is fully executable: exact commands ('entire session current --json', the git rev-parse/log/grep trailer extraction, 'cd <repo> && entire session attach <session-id> --agent <agent> --force'), concrete keep/drop decision rules per candidate, and exact output formats including the preview table and failure messages. This matches anchor 5's copy-paste-ready commands covering the common cases. | 5 / 5 |
Workflow Clarity | A clear five-step sequence with strong validation for a destructive batch operation: the STOP rules mandate preview-before-attach, Step 4 requires explicit user confirmation, Step 5 captures exit status per repo, and the Failure modes section defines recovery (no auto-retry, idempotent re-runs, surfaced stderr). This satisfies the batch/destructive feedback-loop requirement at anchor 5. | 5 / 5 |
Progressive Disclosure | No bundle files exist (references/, scripts/, assets/ are absent) and the body references none, so everything lives in a single well-sectioned file (~137 lines). All of the inline content is runtime-needed procedure rather than material that clearly belongs in a separate file, so structure is good with only minor organization gains possible (e.g. splitting the runtime-transcript lookup details into a reference) — anchor 4. | 4 / 5 |
Total | 18 / 20 Passed |