Skip to content

dongshuyan/compass-skills / skills/run-history-skill-builder

run-history-skill-builder

Verified

Turn a completed task, browser flow, artifact pipeline, failure-recovery trace, or repeatedly refined workflow into a new reusable skill package or a reviewed skill-design plan. Use when the user asks to make a new skill from real run history, extract a reusable workflow from conversation/logs/files, summarize lessons into a new skill, or produce a plan before writing files. Do not use to upgrade an existing skill or to execute the business workflow itself.

evals/evals.json

View on GitHub
{
  "skill_name": "run-history-skill-builder",
  "evals": [
    {
      "id": 1,
      "prompt": "Turn the browser-based upload workflow we just stabilized into a new skill under this repository's skills directory.",
      "expected_output": "A new skill package with trigger wording, a state machine, submission gates, failure-recovery examples, privacy boundaries, and validation results.",
      "files": [],
      "expectations": [
        "The skill package captures reusable state gates instead of a prose summary only.",
        "The package separates public metadata from account-private data and credentials.",
        "The response reports actual validation commands and results."
      ]
    },
    {
      "id": 2,
      "prompt": "Give me a reviewed design plan for a new skill based on this document-conversion workflow, but do not write files yet.",
      "expected_output": "A plan-only response that names candidate files, structure owners, validation gates, and remaining assumptions without writing files.",
      "files": [],
      "expectations": [
        "The response clearly says that no files were created or modified.",
        "The plan identifies which resources are necessary and which are intentionally omitted.",
        "The plan includes validation and acceptance checks."
      ]
    },
    {
      "id": 3,
      "prompt": "We finally stabilized a local DOCX to PDF pipeline. Extract it into a reusable skill.",
      "expected_output": "A new skill that reconstructs the workflow, captures deterministic checks, and packages the recurring failure recovery.",
      "files": [],
      "expectations": [
        "Evidence collection happens before package writing.",
        "The resulting skill includes deterministic validation or script hooks.",
        "The final response distinguishes verified facts from remaining assumptions."
      ]
    },
    {
      "id": 4,
      "prompt": "Search similar public skills in a temporary directory, study their structure, then create a new skill from our run history.",
      "expected_output": "A new skill package plus a source log that records which public references were actually read and which patterns were adopted or rejected.",
      "files": [],
      "expectations": [
        "The response records real sources instead of citing uninspected repositories.",
        "The package adopts patterns, not undocumented claims or marketing text.",
        "The temporary research directory is not shipped as part of the final skill."
      ]
    },
    {
      "id": 5,
      "prompt": "Execute the browser upload flow again for me.",
      "expected_output": "This request should not use run-history-skill-builder because it asks to execute a business workflow, not to create a skill.",
      "files": [],
      "expectations": [
        "The response recognizes the request as a non-trigger.",
        "The response redirects to the relevant business workflow or asks whether the user wants a skill package."
      ]
    },
    {
      "id": 6,
      "prompt": "Based on this failure, improve the existing release-publisher skill.",
      "expected_output": "An upgrade handoff for run-history-skill-upgrader, not a new skill package and not direct edits to the existing skill.",
      "files": [],
      "expectations": [
        "The response recognizes that the request targets an existing skill.",
        "The response produces a clean handoff with evidence scope, structure owners, and validation advice.",
        "The response does not create a new skill package."
      ]
    },
    {
      "id": 7,
      "prompt": "Create a skill that skips questions when context seems sufficient and executes directly when risk looks low.",
      "expected_output": "A rewritten design that removes weak subjective routes and replaces them with user confirmation or machine-verifiable gates.",
      "files": [],
      "expectations": [
        "Weak branches such as 'seems sufficient' or 'looks low risk' are removed or hardened.",
        "User-owned decisions remain with the user.",
        "Machine-verifiable checks move to scripts, tests, schema checks, diffs, or validators."
      ]
    }
  ]
}
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 322kUpdated Claude CodeCodexVerifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 118kUpdated Claude CodeCodexVerifiedInstall