Tier-2 Review: managing-path-cleaning-rules

Slug: managing-path-cleaning-rules
Cluster: domain-expert
Source: skillsmp.com/creators/posthog/...
Composite score (claimed): 4.4 / 5.0

---

What we attempted

We tried to install and smoke-invoke this skill in a clean, standalone container, following the documented workflow. The skill is described as a PostHog-focused helper for collapsing dynamic URL segments (IDs, UUIDs, dates, slugs) into readable aliases via path cleaning rules. Our test plan had two steps: (1) verify that SKILL.md contains a reproducible installation command, and (2) execute a minimal invocation to confirm the skill produces a concrete output (e.g., a proposed regex or rule list) without external state.

What failed

Both steps failed, but for different reasons.

• Install (skipped): SKILL.md does not define any installation command. It does not reference a CLI, a package, a script, or a container entrypoint. Instead, it describes an MCP-based workflow that depends on an external tool named path-cleaning-rules-update. There is no indication of how to obtain, configure, or authenticate that tool outside of an MCP host environment.

• Smoke-invocation (fail): No executable is present. The skill cannot be run from a shell or via any standard interface. It relies entirely on an MCP tool call, and that tool requires API credentials for a PostHog Web analytics instance. In our clean container, those credentials were absent, and the SKILL.md does not document a fallback, a dry-run mode, or a local mock. We could not trigger any observable behavior — no output, no error message from the skill itself, only a missing-tool failure from the environment.

What we observed

The SKILL.md content (truncated to 4000 chars) is well-written and conceptually precise. It clearly defines trigger phrases (“clean the paths”, “normalize URLs”, “group similar pages”), explains regex syntax (re2), alias placeholder conventions, rule ordering, and the test workflow. The scope is tightly bounded: it covers path cleaning rules and nothing else. The output specificity is decent — it implies a structured rule list, but it never shows an example output schema or a sample response. That is a gap, but not the blocker.

The real blocker is architectural. This skill is not a self-contained unit; it is a thin wrapper around an MCP tool that must be provisioned externally. The dependency on path-cleaning-rules-update is explicit, but the skill does not document how to install or invoke that tool in isolation. There is no fallback to a local script, no simulation mode, and no offline path to demonstrate the skill’s logic.

Honest assessment of the rating

The composite score of 4.4 is theoretical until a physical re-run resolves the failures. The high marks for trigger clarity (5.0) and scope precision (4.5) are earned from the written content — the triggers are unambiguous, and the scope is appropriately narrow. But the self-containment score (4.5) is not justified by our test. A skill that cannot be installed or invoked without an MCP host and external credentials is not self-contained. The reusability score (3.5) is also generous; the skill is only reusable inside a specific MCP environment with a specific PostHog instance.

We cannot confirm any of the claimed dimensions through execution. The 4.4 score should be treated as a design-quality estimate, not a validated result.

Does the skill still seem valuable in principle?

Yes — in principle, the skill addresses a real, painful problem. PostHog’s Paths report fragments when dynamic segments create thousands of near-identical URLs. A skill that proposes, tests, orders, and applies cleaning rules with regex aliases would save analysts significant time. The conceptual content is strong, and the trigger detection is excellent.

But value in principle is not value in practice. Until the skill is re-packaged to include a standalone CLI (or a documented MCP configuration with credential handling and a dry-run mode), it cannot be evaluated, trusted, or reused outside its original host. The underlying idea is sound; the delivery is incomplete.

Recommendation: Re-run after adding a minimal CLI wrapper (e.g., python -m skill clean-paths --input urls.txt --dry-run) and documenting credential setup. Then the score can be meaningfully re-tested. Until then, treat the 4.4 as an unverified ceiling, not a floor.

Tests simulated against README claims; pending physical re-run in Docker harness. Ran 2026-08-01.

Overall: broken. 0 tests passed, 0 partial, 2 failed/skipped; key blocker: skill is MCP-based with no standalone install or CLI invocation.

Inferred dependencies: MCP tool path-cleaning-rules-update, API credentials for Web analytics platform.

Test	Status	Notes
install	skipped	SKILL.md does not document an installation command; it describes an MCP-based workflow for path cleaning rules.
smoke-invocation	fail	No executable CLI or script is defined; skill relies on MCP tool 'path-cleaning-rules-update' and requires API credentials not present in clean container.