Content
56%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 curated second half is a genuinely good skill: concrete API commands, a score-based triage table, an explicit verification step, and a full quick-start script. It is undermined by the auto-generated first half — an irrelevant, partially duplicated omniroute CLI manual with at least one truncated sentence — which wastes most of the token budget and buries the actual workflow. Removing or extracting that reference and adding the missing clone/copy commands would make this a strong skill.
Suggestions
Delete the generated 'Subcommands' reference (or move it to a `references/cli-reference.md` linked from a short 'CLI basics' section) — it documents `autostart`, `env`, `setup`, and `update` commands unrelated to skill collection and duplicates `config get`/`set`/`list`, which is the main driver of the low conciseness score.
Fix the broken generated content: the truncated 'Generate OpenCode config (alias for ' sentence and the duplicate `config get`, `config set`, and `config list` sections.
Make the final install step fully executable by showing the actual GitHub clone + `cp SKILL.md` commands the agent must run after the install-planning POST, and add a short error-recovery note for failed API calls or unwritable target directories.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Roughly the first 60% of the body is an auto-generated omniroute CLI subcommand reference (`autostart`, `config`, `env`, `setup`, `update`) that is unrelated to the skill's stated purpose, with duplicated sections (`config get`, `config set`, `config list` each documented twice) and a truncated sentence ('Generate OpenCode config (alias for ' — nothing follows). This is noticeably verbose padding beyond the genuinely lean curated workflow section. Not 3 because the irrelevant bulk is far more than 'some unnecessary explanation'; not 1 because the curated half is efficient and free of concept over-explanation. | 2 / 5 |
Actionability | The curated section provides executable curl commands with headers and JSON payloads, a realistic example response, a copy-paste-ready quick-start bash script, and verification `ls` commands — concrete and mostly complete. Not 5 because the actual file sync ('cloning from GitHub and copying SKILL.md... done by the agent using standard `curl` + `cp` commands') is described but never shown as executable commands, leaving the final install step to be improvised. | 4 / 5 |
Workflow Clarity | The curated workflow is clearly sequenced as Step 1 Detect → Step 2 Review (with a score-prioritization table) → Step 3 Install (with target paths) → Step 4 Verify (explicit `ls` checks and re-running detection), so most checkpoints are present. Not 5 because there is no error-recovery feedback loop (e.g., what to do when the API call fails or a target tool is not writable) and no checkpoint between install planning and execution; not 3 because explicit verification does exist. | 4 / 5 |
Progressive Disclosure | The body has section headers and a well-organized curated half, but the ~265-line generated CLI subcommand reference is bulk content that clearly belongs in a separate reference file and is inlined in SKILL.md (with duplicated entries), matching 'content that should be separate is inline'. The only external pointer (`docs/frameworks/SKILLS.md`) is a passing mention rather than a clearly signaled reference. Not 2 because there is real structure and navigation is possible; not 4 because the largest block of content is misplaced rather than having only minor organization gaps. | 3 / 5 |
Total | 13 / 20 Passed |