Content
57%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, genuinely domain-informative body with concrete tool names, parameters, and a useful quick-reference table, but it is held back by duplicated pitfall content, a filler closing section, missing validation for batch/destructive operations, and an all-inline layout with no reference files. Consolidating redundancy and adding verification steps would lift it substantially.
Suggestions
Add post-operation validation to destructive/batch workflows: after creating a flag, verify via POSTHOG_RETRIEVE_FEATURE_FLAG_DETAILS before enabling; after batch event capture, confirm ingestion before reporting success.
Deduplicate pitfalls — merge the 'Known Pitfalls' section into the per-workflow 'Pitfalls' blocks (the '$' prefix, kebab-case, and flag-uniqueness rules each appear twice) and delete the vacuous 'When to Use' section.
Split the per-tool parameter detail and quick-reference table into a references/ file (e.g., TOOLS.md), keeping SKILL.md as an overview with well-signaled one-level-deep links.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Most content is genuinely useful domain knowledge (tool names, parameters, pitfalls), but there is clear redundancy: the '$' prefix rule, kebab-case, and flag-uniqueness pitfalls each appear twice (per-workflow 'Pitfalls' and again in 'Known Pitfalls'), pagination is restated three times in one list, and the closing 'When to Use' section is pure filler ('This skill is applicable to execute the workflow or actions described in the overview.'). This fits 'mostly efficient but could be tightened'; it does not teach concepts Claude already knows (level 2) but has more duplication than level 4 allows. | 3 / 5 |
Actionability | Concrete tool identifiers, required/optional parameter lists per workflow, a copy-paste-ready JSON example for flag targeting, and a quick-reference table give mostly executable guidance. Minor gaps: no full example call payloads with actual argument values (e.g., a sample POSTHOG_CAPTURE_EVENT invocation), keeping it below 'fully executable, copy-paste ready' level 5; it clearly exceeds the pseudocode/incomplete level 3. | 4 / 5 |
Workflow Clarity | Sequences are clearly listed and setup has explicit checkpoints ('Confirm connection status shows ACTIVE before running any workflows'), but workflows involving batch and destructive operations lack validation steps: batch event capture has no verification despite noted async ingestion, and flag creation/deletion ('Deleting a flag is permanent', 'active: true immediately enables it') has no post-operation verification. Per the guidelines, missing validation in destructive/batch workflows caps workflow_clarity at 3; the presence of the well-checked setup sequence keeps it above level 2. | 3 / 5 |
Progressive Disclosure | The body has clear section structure (workflows, common patterns, quick reference), but it is a ~220-line monolith with no bundle files at all: the per-tool API reference (parameter lists, pitfalls, and the task table) is inline content that belongs in a separate reference file, leaving SKILL.md as a full API dump rather than an overview. This matches 'some structure but content that should be separate is inline'; it is better organized than the header-less level-2 wall of text but lacks the file split of level 4. | 3 / 5 |
Total | 13 / 20 Passed |