Content
100%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A well-structured capability skill body that pairs executable CLI examples with clear decision guidance, keeps the dense reference material in one-level-deep bundle files, and avoids restating concepts Claude already knows. The 'when not to' gate and explicit disposition options give it real workflow clarity even though it intentionally avoids a fixed procedure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean and assumes Claude's competence — no explaining of basic CLI or memory concepts; the prose present (multi-platform session-root paths, 'tool not a ceremony') is project-specific knowledge or judgment guidance that earns its tokens rather than filler. | 3 / 3 |
Actionability | Provides concrete, copy-paste-ready bash commands with real flags — 'trellis mem search "<keyword>"', 'extract <session-id> --phase brainstorm', 'context <session-id> --turns 3 --around 2', 'list --cwd <project-path>' — plus concrete phase-boundary semantics. | 3 / 3 |
Workflow Clarity | Explicitly framed as a capability skill rather than a fixed workflow; the 'When to reach' → 'When NOT to reach' → 'What to do with results' decision flow includes an explicit gate ('If none of these apply, don't call `mem`') and a 'When NOT to' checkpoint, which is appropriate for a non-procedural skill. | 3 / 3 |
Progressive Disclosure | SKILL.md is an overview; the full CLI reference and triggering patterns are split into two well-signaled one-level-deep bundle files (references/cli-quick-reference.md, references/triggering-patterns.md), both verified present with no nested references. | 3 / 3 |
Total | 12 / 12 Passed |