Content
67%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.
The body delivers concrete, sequenced, safety-conscious operational guidance with real validation gates (dry-run sha256 approval, contracts verification, idempotent no-op semantics) and well-organized sections. Its weaknesses are moderate redundancy in the boundary-policy prose, a pinned commit/date that belongs in a reference, and undefined shell variables that keep the examples from being fully copy-paste ready.
Suggestions
Trim the repeated prohibition statements in 'Why execution stops here' into a single compact list, and move the pinned upstream commit hash and 'checked 2026-10-17' date into the referenced wiki documentation or a deprecated/notes section to avoid time-sensitive content penalization.
Define how $VAULT, $GENERATED_AT, and $OPERATION_ID are obtained (e.g., one line each) so the capture-apply commands become copy-paste ready.
Briefly state the expected action when 'contracts --verify' or a capture apply fails, converting the implicit error paths into explicit feedback loops.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and operational with no basic-concept padding, but the 'Why execution stops here' section repeats boundary statements ('never run it merely because it was found', 'Never relabel configured as verified', 'Do not execute… Do not silently substitute…') that could be tightened. The pinned commit hash 'checked 2026-10-17' is time-sensitive information not placed in a deprecated/old-patterns section, which the guidelines penalize. Matches anchor 3: mostly efficient but could be tightened. | 3 / 5 |
Actionability | Concrete, mostly executable bash blocks are given: 'python3 "$CORE" capture external-plan url "HTTPS_URL"', 'capture plan --vault "$VAULT"', and the dry-run-then-'--approved-plan-sha256 "$APPROVAL_SHA256" --apply' sequence, with a comment explaining how to set APPROVAL_SHA256. Anchor 4 rather than 5 because $VAULT, $GENERATED_AT, and $OPERATION_ID are used but never defined, so it is not fully copy-paste ready. | 4 / 5 |
Workflow Clarity | Both workflows are clearly sequenced with real checkpoints: 'test -f "$CORE"', 'contracts --verify', a dry-run plan whose 'approved_plan_sha256' must be reviewed before applying, 'Inspect the actual result and record extraction limitations', and the idempotency rule 'A repeated identical payload is a no-op'. Anchor 4 rather than 5 because error-recovery paths (what to do when verification or apply fails) are left implicit. | 4 / 5 |
Progressive Disclosure | Sections are well organized (product root resolution, inert URL plan, execution-stop rationale, filesystem capture), and the single reference 'Use [operation-transactions.md](../wiki/references/operation-transactions.md) for custom capture drafts' is clearly signaled and one level deep. Anchor 4 rather than 5 because the reference resolves outside the skill bundle and the ~18-line inline policy rationale in 'Why execution stops here' could live in a reference file to keep SKILL.md lean; no bundle files (references/, scripts/, assets/) exist in this package. | 4 / 5 |
Total | 15 / 20 Passed |