Tessl • The package manager for agent skills

SkillPublic

feature-usage-feed

Set up an LLM-judge evaluation that extracts canonical use cases for a PostHog feature at scale and streams the results to a Slack channel as a live feed. Use when someone wants to understand how users are actually using a specific AI/LLM-powered feature in production — what they're investigating, what questions they're trying to answer, and what patterns surface — without manually reading hundreds of traces. Assumes the feature emits `$ai_generation` and `$ai_evaluation` events with `$session_id` linkage to the trigger user's recording (the standard setup post the session-summary linkage PRs).

Quality

87%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

—

The risk profile of this skill

Quality

Content

85%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A rich, highly executable body that scores top marks on actionability and workflow clarity thanks to concrete payloads, validation checkpoints, and feedback loops. The main weakness is progressive disclosure: at ~450 lines it is a self-contained monolith with no reference files, and some material could be split out.

Suggestions

Split the 'Worked example A' and 'Worked example B' narratives into a references/ file (e.g. WORKED_EXAMPLES.md) and link to them from the workflow steps, leaving the main body as an overview.

Move the 'Validating signal quality after launch' section into a separate reference file and reference it from Step 7 to keep the core workflow terse.

Tighten the intro paragraph and the worked-example prose ('Reading them by hand is not feasible') to trim tokens without losing the load-bearing gotchas.

Dimension	Reasoning	Score
Conciseness	Assumes Claude's competence (no explanations of what LLM judges, traces, or Slack are) and is mostly load-bearing operational gotchas, though the two lengthy worked-example narratives could be trimmed.	4 / 5
Actionability	Copy-paste-ready HogQL queries, complete JSON payloads for every posthog:llma-evaluation-* tool, and a full Slack block-kit template with clearly marked user-substituted placeholders cover the common cases end to end.	5 / 5
Workflow Clarity	A clear 7-step sequence with explicit validation checkpoints (volume/coverage checks, 3-5 trace iteration, synthetic-vs-real test panel, end-to-end production verify) and feedback loops including a failure-mode table and a recovery path when 'View Trigger Session' lands on the homepage.	5 / 5
Progressive Disclosure	Well-organized into named sections but it is a ~450-line monolith with no bundle files, and content that could live in separate references (the worked examples, the signal-quality validation guide) is inlined rather than split out.	3 / 5
	Total	17 / 20 Passed

Description

83%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong, third-person description that explicitly states what the skill does and when to use it with natural trigger language. It is concrete and well-scoped to a PostHog AI-observability niche, with only minor gaps in action breadth and overlap risk against sibling trace-eval skills.

Dimension	Reasoning	Score
Specificity	Lists several concrete actions ('Set up an LLM-judge evaluation that extracts canonical use cases... and streams the results to a Slack channel as a live feed') but stops just short of the fully comprehensive multi-action coverage of the score-5 anchor.	4 / 5
Completeness	Explicitly answers both what ('Set up an LLM-judge evaluation that extracts canonical use cases... and streams the results to a Slack channel as a live feed') and when ('Use when someone wants to understand how users are actually using...') with concrete trigger phrases.	5 / 5
Trigger Term Quality	Strong natural phrasing ('Use when someone wants to understand how users are actually using a specific AI/LLM-powered feature') with synonyms like 'canonical use cases' and 'patterns surface'; a few natural terms are absent but there are no file extensions to cover for this non-file skill.	4 / 5
Distinctiveness Conflict Risk	Clear PostHog-specific niche anchored on $ai_generation/$ai_evaluation events with $session_id linkage, but there is residual overlap risk with the closely related exploring-llm-traces / exploring-llm-evaluations skills that the body has to redirect away from.	4 / 5
	Total	17 / 20 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: PostHog/posthog
Path: products/ai_observability/skills/feature-usage-feed/SKILL.md
Revision: 130f3a1

Reviewed: 29 days ago

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.

feature-usage-feed

Table of Contents

Is this your skill?