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.
This is a high-quality reference skill: it documents a machine-readable contract (typed exits, JSON envelopes, confirmation gates) with fully executable commands and explicit validation and recovery loops for every risky path. The main improvement levers are trimming the couple of cross-section repetitions and making the offloaded chat-vs-search decision table available within the bundle.
Suggestions
State the `auth token` raw-token default once (§3's output-contract exception is the natural home) instead of twice in §1 and §3.
Ship the chat-vs-session-ask-vs-search decision table (or a bundle reference file backing the `weknora-rag-search` pointer) so readers of this skill alone can resolve the CLI's most common confusion.
Condense the §7a tool-approval exit-10 note to a single cross-reference to §5 rather than restating the rule.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and almost entirely non-obvious CLI-specific contract (exit-code semantics, typed error codes, batch envelope shapes) with essentially no filler Claude would already know. It misses the 5 anchor because of small redundancies — the `auth token` raw-token exception is stated in §1 and repeated as an exception in §3, and the exit-10 approval rule is restated in §7a — but is well above the 3 anchor's 'some unnecessary explanation'. | 4 / 5 |
Actionability | Every section is copy-paste ready: concrete command lines ("weknora profile add prod --host https://kb.example.com --use", "echo \"$WEKNORA_API_KEY\" | weknora auth login --with-token"), an exit-code table with per-code agent actions, a realistic JSON error envelope, and a working chaining example (KB=$(weknora kb create ... --jq '.data.id')). This matches the fully-executable anchor. | 5 / 5 |
Workflow Clarity | Multi-step processes are explicitly sequenced with validation checkpoints: the auth section marks order as 'do this first — order matters', verifies with `auth status`, uses `kb status`/`kb check` for `retrieval_ready`, prescribes `--dry-run` before mutations, and defines exit-code-driven recovery (re-auth on exit 3, backoff on 6/7, user approval on exit 10) plus batch failure inspection. Destructive and batch operations all have feedback loops, satisfying the cap. | 5 / 5 |
Progressive Disclosure | With no bundle files present, the single-file organization is strong: numbered, well-labeled sections (auth, KB selection, output contract, exit codes, destructive writes, chaining, command map, self-help) plus a 'Common mistakes' table make navigation easy, and the one deep-dive (chat vs session ask vs search) is deliberately offloaded to a sibling skill. It falls short of 5 because the pointer targets `weknora-rag-search` outside this bundle, so the decision table is not actually reachable from what ships here, and some detail (e.g., the full per-resource command map) is inline rather than split out. | 4 / 5 |
Total | 18 / 20 Passed |