Content
75%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 well-structured, instruction-only dispatcher skill: concrete routing commands with defaults, clear validation and error-recovery guidance, and good deferral of setup detail to external docs. The recurring cost is redundancy in the Security section and the absence of a worked example of an actual HogQL query through the exec tool.
Suggestions
Consolidate the Security section: state the read-only requirement once (the bolded recommendation plus its rationale) and drop the repeated 'never create, update or delete' paragraph, since the no-credential point is already made in the frontmatter and intro.
Add one short example of delegating a query (e.g., the HogQL SELECT for per-day organic visits via `call execute-sql`) so the common case is copy-paste ready.
Trim the cross-skill spelling digression ('mcp__posthog__exec ... or mcp__PostHog__exec otherwise. Either spelling counts.') to a single line, e.g. 'Accept either `mcp__posthog__exec` or `mcp__PostHog__exec`.'
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient — a tight routing table, brief prerequisites, and pointed error handling — with no explanations of concepts Claude already knows. The Security section states the read-only rule three times and the no-credential point appears in the frontmatter, intro, and Security section, which is trimmable redundancy keeping it below anchor 5. | 4 / 5 |
Actionability | The routing table gives concrete commands with per-command purposes, specifies the exec path ("HogQL queries through the MCP `exec` tool (`call execute-sql`)"), defaults ("days (default 28) and limit (default 50)"), and stepwise error handling. It stops short of anchor 5 because no example HogQL query or exec invocation is shown, leaving a minor gap for the common query cases. | 4 / 5 |
Workflow Clarity | Dispatch is unambiguous per command, with a dedicated validation step (`/seo posthog check` confirms MCP connectivity and `$pageview` events) and an error-recovery checkpoint (verify "project, window, and that `$referring_domain` is captured before reporting zero organic traffic"). It does not reach anchor 5 because the cross-skill audit delegation flow has no explicit end-to-end sequence with feedback loops. | 4 / 5 |
Progressive Disclosure | Sections are well organized, appropriately sized for a SKILL.md overview, and setup detail is correctly deferred to clearly signaled one-level-deep pointers ("extensions/posthog/docs/POSTHOG-SETUP.md"). It falls short of anchor 5 because those pointers reference paths outside the skill bundle rather than local reference files, and the ~75-line body exceeds the under-50-line simple-skill threshold the guideline cites for a top score. | 4 / 5 |
Total | 16 / 20 Passed |