agentsclimarketplace
Sign in

Weekly test pulse

Skill intelligems-io/intelligems-mcp-skills/skills/weekly-test-pulse

Use when a brand operator or agency asks for a quick weekly status report on running Intelligems tests and recommended next actions.From its SKILL.md

Install
npx -y skills add intelligems-io/intelligems-mcp-skills --skill weekly-test-pulse

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its file declares

Copied from the file, not written here

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

1.7 KB, 335 tokens by cl100k_base, as published. Nobody here has run it

Weekly Test Pulse

Use this skill for a concise status report on tests that are currently running.

Steps

  1. Resolve the organization. If ambiguous, use list_organizations and ask for the target.
  2. Find running tests with search_experiments or search_experiences using status: "started".
  3. For each running test, fetch details with get_experience and metrics with analyze_experience.
  4. Use get_experience_metrics_config when you need the correct analytics view type or primary metric.
  5. Group tests into: likely winner, likely loser, inconclusive, needs more data, and needs attention.

Watchouts

  • Do not recommend ending a test solely from early conversion lift.
  • Check RPV, GPV, net revenue, AOV, and confidence when available.
  • Call out tests with low traffic, uneven traffic allocation, missing profit metrics, or suspicious segment concentration.
  • If a lifecycle action is needed, recommend it but do not run experience_action without explicit approval.

Data Safety

Use only the user's authenticated MCP or API context. Do not save raw responses, exports, graph URLs, order data, customer data, or private store identifiers to the repository. Summarize findings and include experience IDs only when needed for the user's workflow.

Output

Return:

  1. One-paragraph pulse summary.
  2. Table of running tests with status, leading variation, metric direction, confidence, and recommendation.
  3. Risks or anomalies.
  4. Exact next actions for the coming week.
  5. Source appendix with organization, date range, and tool calls.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 326,736. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.