Content
71%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 is well-organized, efficient, and honest about evidence boundaries, with a clear workflow and a properly signaled one-level reference structure. Its main weakness is actionability: it tells Claude what to identify but never shows what a good hook, outline, or script output actually looks like, and several constraint lines are generic boilerplate irrelevant to a writing task.
Suggestions
Add 1-2 concrete output examples (e.g., a worked hook with promise/tension/payoff labeled, or a short outline skeleton) so the Examples section demonstrates the artifact rather than only listing user prompts.
Replace the generic repo/destructive-command constraint boilerplate with constraints specific to video scripting (e.g., how to handle unverified metrics or client-provided claims).
Make the validation checkpoints checkable: define what 'the payoff answers the hook' means (e.g., each hook promise maps to a specific script section) so step 5 is a test rather than a vibe.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly lean, well-sectioned bullet lists with no explanations of concepts Claude already knows. However, parts of Constraints ('Make repo-owned changes only after confirming the target path', 'Do not run destructive commands') and 'Start with 2-3 focused surfaces before expanding scope' read as generic template boilerplate that does not earn its place in a video-scripting skill. Anchor 4 ('minor instances that could be trimmed') fits better than anchor 5. | 4 / 5 |
Actionability | The workflow gives directive steps ('Identify the opening promise, viewer tension, and concrete payoff') and concrete input/output contracts, but no concrete output artifacts: no hook structures or formulas, no example hook or outline format, and the Examples section shows sample user prompts rather than example outputs. This lands on anchor 3 ('some concrete guidance but incomplete; missing key details') rather than 4, which expects mostly executable guidance. | 3 / 5 |
Workflow Clarity | A clear six-step sequence with explicit checkpoints (step 5's payoff check, step 6's fail-fast) and a dedicated Validation section. Not anchor 5 because the checkpoints are qualitative with undefined criteria — nothing specifies what makes a payoff 'answer' the hook or how to judge audience fit — leaving minor validation gaps characteristic of anchor 4. | 4 / 5 |
Progressive Disclosure | A ~75-line body organized into clear sections, with a dedicated Progressive Disclosure section that signals the active reference files (references/contract.yaml, references/evals.yaml, references/task-profile.json — all verified to exist) with their purposes, and points to archived long-form context by path. References are one level deep and well-signaled, matching the anchor-5 'clear overview with well-signaled one-level-deep references' example. | 5 / 5 |
Total | 16 / 20 Passed |