Content
78%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.
A well-structured capability skill: executable command coverage of the common cases, explicit when/when-not guidance, and clean one-level-deep reference split. The main cost is token weight from the inlined agent-path catalog and meta-commentary that the CLI reference could carry.
Suggestions
Move the per-agent storage path catalog (Claude, Codex, Devin, Grok, OpenCode, Pi, ZCode paths) out of the 'What trellis mem is' section into references/cli-quick-reference.md, keeping only the one-sentence summary that mem indexes multiple agents' local logs.
Trim the meta-commentary sentences ('It is intentionally a capability skill, not a workflow', 'Forcing every recall into a fixed file makes the file grow into noise') to a single clause — the surrounding bullet structure already conveys the point.
In 'How to call it', convert the commented code block's sequence guidance into a short ordered decision path (start with search → if session id unknown use list/projects → extract or context → choose a disposition) to make the implicit workflow explicit.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The when-to/when-not-to sections are tight, but the agent-storage path catalog (exact paths for Devin, Grok, OpenCode, Pi, ZCode) duplicates material that belongs in the CLI reference, and meta-commentary like 'It is intentionally a capability skill, not a workflow' and 'Forcing every recall into a fixed file makes the file grow into noise' could be trimmed — matching 'mostly efficient but includes some unnecessary explanation'. | 3 / 5 |
Actionability | Copy-paste-ready commands covering the stated 80% cases — 'trellis mem search "<keyword>"', 'trellis mem extract <session-id> --phase brainstorm', 'trellis mem context <session-id> --turns 3 --around 2', 'trellis mem list --cwd <project-path>' — with explanatory comments and phase-slicing semantics, matching the fully-executable top anchor. | 5 / 5 |
Workflow Clarity | A clear decision workflow (when to reach for it, when not to, what to do with output) with an implicit call sequence guided by comments like 'When you do not know the session id yet, start with list + filter'. It is a read-only capability skill so no validation cap applies, but the sequence lacks the explicit step numbering and checkpoints of the top anchor. | 4 / 5 |
Progressive Disclosure | The ~75-line overview keeps only the 80% case inline and clearly signals two real one-level-deep references — 'Full CLI reference is in `references/cli-quick-reference.md`' and '`references/triggering-patterns.md` lists more verbatim user phrasings' — both of which exist as actual bundle files, matching the top anchor. | 5 / 5 |
Total | 17 / 20 Passed |