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.
A well-engineered operational skill: concrete commands with locking semantics, privacy filtering, manifest-based incremental processing, and explicit validation and failure handling for both the ingest and the QMD refresh. The main improvable areas are minor redundancy between the layout tree and the ranked source list, and a few template placeholders plus unverifiable external reference paths.
Suggestions
Merge the 'Key data sources ranked by value' list into the data-layout tree (e.g., annotate each path with its value rank) to remove the duplicated walkthrough of the same five sources.
Move the manifest JSON schema and the QMD refresh procedure into a reference file (e.g., references/qmd-refresh.md), keeping SKILL.md to a short pointer plus the status-vocabulary list.
Replace template placeholders like `lifecycle_changed: <obsidian-wiki memory now --date>` with a literal example invocation so the field is copy-paste ready.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense operational instruction that assumes competence — no padding explaining what JSONL or Obsidian is, and it states rules tersely ("Skip credentials/ entirely. Skip agents/*/agent/models.json"). Below anchor 5 because the data-layout tree and the 'Key data sources ranked by value' list cover the same ground twice, and the manifest JSON example includes illustrative counts that could be trimmed. | 4 / 5 |
Actionability | Concrete, copy-paste-ready commands dominate: `obsidian-wiki memory sync OPENCLAW_HISTORY_INGEST memory=<memory> ... --takeaways "..."`, `${QMD_CLI:-qmd} update/embed/ls/get`, plus executable JSON/JSONL/manifest examples. Not anchor 5 because some snippets remain templates — `lifecycle_changed: <obsidian-wiki memory now --date>` and placeholder counts in the manifest block — leaving minor gaps for a first-time executor. | 4 / 5 |
Workflow Clarity | A clearly sequenced 7-step pipeline with explicit checkpoints and feedback loops appropriate to a batch operation: delta classification (New/Modified/Unchanged), "Report a concise delta summary before deep parsing", and the QMD section's validate-then-branch flow ("If the output says vectors are needed or embeddings may be stale, run qmd embed") with explicit failure handling ("If QMD refresh fails, do not roll back the vault changes; report the QMD status separately") and a recorded status vocabulary. The batch-operation validation cap does not apply because validation is present. | 5 / 5 |
Progressive Disclosure | The single bundle reference is well-signaled and real: "See `references/openclaw-data-format.md` for field-level notes and parsing guidance" points to an existing file, and externalized procedure lives in named external skills (llm-wiki/SKILL.md). Below anchor 5 because several in-bundle/external paths (`.skills/llm-wiki/references/MEMORY.md`, `references/` inside llm-wiki) cannot be verified from this bundle and add navigation risk, and the SKILL.md itself is long enough (260 lines) that some inline detail (manifest schema, QMD refresh) could be split into references. | 4 / 5 |
Total | 17 / 20 Passed |