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.
A dense, command-first routing skill that is exceptionally actionable — exact commands, exit codes, output contracts, env vars and error-recovery paths — with a well-verified one-level reference bundle. The residual weaknesses are inline time-sensitive model/date detail that duplicates the capabilities and prices references, and a few pockets of deep operational detail (Grok isolation, key config format) that belong in reference files.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes Claude's competence — no concept explanations, just operational facts in command tables — but it inlines time-sensitive detail (model ids like 'gpt-image-2.5-sunburst' and the 'owner decision 2026-10-06' date) that duplicates references/capabilities.json and references/prices.json and will go stale, which the rubric's time-sensitivity guideline penalizes. It is efficient with minor rot-prone content to trim, matching the level-4 anchor rather than the rot-free level 5. | 4 / 5 |
Actionability | Fully executable, copy-paste-ready guidance: exact commands with full argument sets in need/command tables (e.g. 'route_media.py image --prompt-file <txt> --reference <png> --size 1024x1024 --out-dir <new>'), exact exit codes, the exact JSON output shape, error-line format, environment variable names and config file paths. The common cases (resolve, image, video, route choice, dry-run) are all covered, so it fits the top anchor. | 5 / 5 |
Workflow Clarity | The sequence is clear (forge_doctor once per session, then route order, then generate) with explicit validation checkpoints ('--dry-run (no key is used, nothing is sent or written)') and explicit feedback loops for error recovery ('a route whose account cannot serve the request passes it to the next route; a refusal such as moderation stops', plus 'cli_media.py resume --run <id>, then --adopt (never reruns the CLI)'). This matches 'clear sequence with explicit validation steps; feedback loops for error recovery' rather than the level-4 anchor's 'minor validation gaps'. | 5 / 5 |
Progressive Disclosure | The 70-line body is a clear overview with well-signaled, verified one-level references (route-media.md, api-usage.md, cli-routes.md, provider-survey.md, capabilities.json, prices.json, schemas/), but two minor gaps keep it below the top anchor: deep detail is inlined where a reference would do (the Grok HOME/USERPROFILE/GROK_HOME/GIT_CONFIG_GLOBAL isolation paragraph, the full key config file format), and references/agent-profiles/README.md itself indexes onward to video-agent.md, introducing a second hop on one branch. | 4 / 5 |
Total | 18 / 20 Passed |