Content
77%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 content is highly actionable with a well-sequenced, validation-checkpointed procedure and concrete memory/file targets, but it is somewhat verbose (repeated scope-guard/ECHO framing) and relies on external references that are not present as resolvable bundle files. Instruction-only guidance is appropriately concrete despite the absence of code.
Suggestions
Tighten the intro and scope-guard paragraph to avoid restating frontmatter boundary language, and consolidate ECHO cross-reference jargon into a single concise pointer.
Either ship the referenced files (echo-benchmark.md, platforms/tiktok.md, platforms/youtube.md, skill-contract.md, humanizer-slop.md, SECURITY.md) as a ./references bundle, or replace deep ../../../ paths with verified in-bundle links so progressive disclosure lands cleanly.
Move the long Reference Materials link list into a short, well-signaled section so the SKILL.md reads as an overview pointing one level deep rather than interleaving references in prose.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and free of basic-concept padding, but the intro and scope-guard paragraph repeat frontmatter boundary text and lean heavily on ECHO cross-reference jargon; it could be tightened without losing information, fitting 'mostly efficient but some unnecessary explanation or could be tightened' rather than the 'minor instances' bar for 4. | 3 / 5 |
Actionability | Concrete, executable guidance throughout: an 8-step Procedure with per-row fields ('time window / spoken line / on-screen text / asset keywords'), enumerated named hook families, specific memory paths (memory/social/short-video-scripter/, memory/events/claims.ndjson via registry-events.py), and copy-paste Quick Start prompts — code absence is fine for an instruction-only spec skill. | 5 / 5 |
Workflow Clarity | A clearly sequenced 8-step Procedure with explicit validation checkpoints: NEEDS_INPUT gating in step 1, 'Never present an undated spec as current' in step 5, the disclosure/claims pass in step 6, the slop self-check in step 7, an explicit 'Done when' checklist, and termination rules — feedback loops are present for the gated propose writes. | 5 / 5 |
Progressive Disclosure | Sections and a Reference Materials list provide some structure, but no bundle files exist under ./references, ./scripts, or ./assets and the heavily referenced external paths (../../../references/*.md, ../../../SECURITY.md) do not resolve, leaving references present but not clearly verifiable and the ~85-line body leaning toward inline density rather than a clean one-level-deep overview. | 3 / 5 |
Total | 16 / 20 Passed |