Content
93%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 tight, well-structured skill body that gives concrete tool-level guidance and explicit branching for match resolution, including a confirmation checkpoint before destructive actions. The only minor gap is the absence of an explicit error-recovery feedback loop in the workflow.
Suggestions
Add a brief error-recovery note for the flag-key branch (e.g., what to do when `experiment-get-by-flag-key` returns `ambiguous` beyond picking a candidate) to close the workflow_clarity gap.
Consider an explicit 'verify the resolved ID matches the user's intent before acting' checklist step to strengthen the destructive-action validation checkpoint.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean and efficient: it assumes Claude's knowledge (no preamble about what PostHog or experiments are) and every section earns its place, including the non-obvious flag-key semantics. | 5 / 5 |
Actionability | Fully actionable: names the exact MCP tools (`experiment-list`, `experiment-get-by-flag-key`, `experiment-get`, `experiment-create`), their return fields, and concrete branch handling for one/multiple/no matches with a worked example. | 5 / 5 |
Workflow Clarity | Clear sequence (find → resolve → confirm → get details) with explicit checkpoints (confirm before destructive actions, disambiguate multiple, handle no matches), but no explicit error-recovery feedback loop, leaving a minor validation gap. | 4 / 5 |
Progressive Disclosure | Well-organized single-purpose skill with clear section headers and no need for external bundle files; the structure is easy to navigate as a self-contained overview. | 5 / 5 |
Total | 19 / 20 Passed |