agentsclimarketplace
Sign in

Microsoft clarity

Skill holdenjrussell/ms-clarity-cli/skills/microsoft-clarity

Analyze Microsoft Clarity analytics, session recordings, heatmaps, rage clicks, dead clicks, scroll depth, JavaScript errors, and landing-page UX friction. Use for Clarity MCP setup, warehouse syncs, recording triage, behavior analysis, CRO investigations, or evidence-backed site recommendations for any client or brand.From its SKILL.md

Install
npx -y skills add holdenjrussell/ms-clarity-cli --skill microsoft-clarity

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

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 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.

SKILL.md

3.6 KB, 722 tokens by cl100k_base, as published. Nobody here has run it

Microsoft Clarity

Use the ms-clarity CLI and Microsoft Clarity MCP server to turn aggregate behavior and focused recording samples into evidence-backed UX findings. Keep client identity, credentials, database URLs, schedules, and private session data outside this skill.

Start here

  1. Run ms-clarity doctor before live work.
  2. If setup is incomplete, read MCP and runtime setup.
  3. Check local data before consuming API calls:
ms-clarity warehouse status
  1. Prefer warehouse analysis. Use MCP or a live fetch only for missing slices, fresh one-off investigations, or documentation lookup.
  2. Read warehouse and call budgets before any live fetch.
  3. Read CRO workflow before recording or heatmap analysis.
  4. Read privacy and client isolation before handling replay links or raw payloads.

Tool surfaces

  • MCP tools: query-analytics-dashboard, list-session-recordings, and query-documentation-resources.
  • CLI: ms-clarity.
  • Token environment variable: CLARITY_API_TOKEN, loaded from the owner-only local config by ms-clarity mcp.
  • Warehouse tables: api_responses, metric_rows, recording_samples, and sync_runs under the configured schema.
  • Safe setup wrapper: scripts/configure.sh.
  • MCP wrapper: scripts/mcp-wrapper.sh.
  • Warehouse wrapper: scripts/warehouse.sh.

Default analysis sequence

  1. Confirm the date range, project/client, device scope, and affected pages.
  2. Query aggregate metrics by URL, device, source, medium, campaign, browser, operating system, and country.
  3. Cluster repeated symptoms such as rage clicks, dead clicks, quick backs, low scroll, and JavaScript errors.
  4. Select 5 to 15 representative recordings for a repeated pattern. Do not browse random chronological lists.
  5. Human-check or browser-check representative replays before describing the visible cause.
  6. Use exported heatmap CSV or PNG files only with their page, date, device, and filter context.
  7. Recommend a specific fix or experiment with evidence counts and affected surfaces.

Warehouse commands

ms-clarity warehouse status
ms-clarity warehouse fetch-insights --num-days 3 --plan compact --max-calls 4
ms-clarity warehouse fetch-recordings --days 7 --bucket mobile-rage-dead --max-calls 1
ms-clarity warehouse backfill --days 30 --max-calls 9 --insight-calls 6

Never run a live fetch merely to test installation. MCP discovery, doctor, warehouse initialization, and mocked tests do not consume the Clarity project call budget.

Recommendation contract

Finding:
Evidence:
Affected surface:
Likely cause:
Fix/test:
Priority:
Replay links:

Treat a single recording as an anecdote unless it proves a hard bug. Separate observed behavior from inferred cause.

Safety

  • Never print or paste the Clarity token.
  • Do not expose session-level payloads, IPs, emails, order details, private URLs, or replay links in public output.
  • Keep raw data and generated client reports outside the repository.
  • Treat recommendations as read-only until the client approves site, tracking, or campaign changes.
  • Do not install a scheduler, post a report, or mutate a production surface without explicit approval.

What ships with it: 8 files

7.5 KB alongside SKILL.md, 3 of them executable

agents/

references/

scripts/

Keep looking

Skills are one crate of 326,144. 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.