Content
96%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 exceptionally tight operational skill: every rule is concrete, every failure mode of the CLI has a stated recovery, and the destructive memory-rewrite workflow is fully validated. The only structural nit is that all CLI behavioral detail sits inline in a single ~135-line file rather than being split out.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense, rule-driven prose with zero padding or explanation of concepts Claude already knows — e.g. 'promotion.json, checksums of the two, which you can ignore' — and every token carries a non-obvious operational constraint. | 5 / 5 |
Actionability | Guidance is fully executable: exact commands ('tessl memory pending --topic <topic> --out-dir <dir>', 'tessl memory save --topic <topic> --file <note file> --note-version <version>'), an exact line format ('- <fact> (<who>, <YYYY-MM-DD>)'), the facts.json shape, and named tool behavior for every refusal case. | 5 / 5 |
Workflow Clarity | The situation ladder gives a sequenced decision procedure, and every risky path has an explicit validation/feedback loop — version mismatch re-runs 'pending' into a new folder, a >48 KB save deletes lines per invariant 8 and saves again, and a save that gets no answer is retried with interpretation of the follow-up refusal. This is a destructive/batch rewrite where dropped lines are gone, yet validation is exhaustively specified, so no cap applies. | 5 / 5 |
Progressive Disclosure | Sections are well organized (What done looks like / Invariants / Environment facts / Judgment) with nothing buried, and no bundle files exist to reference. Not 5 because at ~135 lines the inlined CLI reference in 'Environment facts' could arguably live in a separate reference file, exceeding the simple-skill threshold where section organization alone earns a 5. | 4 / 5 |
Total | 19 / 20 Passed |