Content
86%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.
An exemplary reference-style skill: almost entirely executable commands, no filler or basic-concept explanations, explicit guardrails for destructive and trust-sensitive operations, and a well-structured two-file bundle with accurate one-level-deep anchored references. The only improvements are removing the Quick Start duplication with the Managing Servers section and making the implicit multi-step workflows (skill publishing, plugin install/sync) explicit numbered sequences.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dominated by terse, commented command blocks with almost no explanation of concepts Claude already knows, and dense high-value sections (error table, guardrails). It is not a 5 because of duplication — Quick Start repeats `thv list`, `thv status`, `thv stop`, `thv rm` verbatim from Managing Servers — and a few prose passages (AI Plugin Management) could be tightened. | 4 / 5 |
Actionability | Nearly every section consists of copy-paste-ready commands with realistic flags and arguments (e.g., `thv run github --secret GITHUB_TOKEN,target=GITHUB_PERSONAL_ACCESS_TOKEN`, `thv run --from-config ./config.json`, `codex plugin install example-plugin@toolhive`), and the examples cover the common cases across all command families, matching the 5 anchor exactly. | 5 / 5 |
Workflow Clarity | Sequences are clear within each task area (install → list → info → uninstall; validate → build → push), prerequisites are stated up front ("Requires `thv serve` to be running"), and validation/recovery exists (`thv skill validate`, `thv runtime check`, `ai-plugin sync --check`, `upgrade --preview`, the Error Handling table, and a guardrail requiring confirmation before destructive operations). It falls short of 5 because it is a command reference where multi-step workflows are implicit orderings rather than explicit step-by-step procedures with checkpoints. | 4 / 5 |
Progressive Disclosure | SKILL.md is a clean overview that pushes full flag documentation to references/COMMANDS.md and detailed usage to references/EXAMPLES.md, both of which exist, are well-organized by sectioned anchors that all resolve (e.g., `#thv-run`, `#skill-commands`, `#ai-plugin-commands`), and contain no further nested references — exactly one level deep, clearly signaled per section and in a See Also list, matching the 5 anchor. | 5 / 5 |
Total | 18 / 20 Passed |