dongshuyan/compass-skills / skills/run-history-skill-builder
run-history-skill-builder
VerifiedTurn 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.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 7.2 KB |
| agents/openai.yaml | config | 347 B |
| evals/evals.json | config | 4.1 KB |
| references/examples.md | markdown | 2.7 KB |
| references/history-mining.md | markdown | 3.3 KB |
| references/open-source-pattern-mining.md | markdown | 1.9 KB |
| references/self-repair-and-evals.md | markdown | 3.0 KB |
| references/skill-design-protocol.md | markdown | 2.6 KB |
| scripts/test_validate_skill_package.py | script | 5.1 KB |
| scripts/validate_skill_package.py | script | 13.1 KB |
references/history-mining.md
View on GitHub# History Mining
Read this file when you need to reconstruct what really happened before packaging it into a new skill.
## Source Priority
1. Current visible conversation and direct user corrections.
2. User-provided files, logs, screenshots, artifacts, and transcript paths.
3. Current workspace files, diffs, tests, generated outputs, and terminal evidence.
4. Public references used only for packaging patterns, not as proof that the local workflow succeeded.
## What To Extract
- user goal and acceptance criteria;
- true inputs, outputs, and side effects;
- successful path;
- failed attempts, blockers, and misjudgments;
- fixes that actually worked;
- approval gates and stop conditions;
- observable success evidence.
## Evidence Sheet
For each candidate rule, record:
- `observation`: what happened;
- `source_type`: conversation, file, log, artifact, browser, diff, test, or public reference;
- `confidence`: high / medium / low;
- `invariant`: the reusable part that belongs in a skill;
- `local_detail`: the path, account, date, or one-off circumstance that should not be released as a rule.
## Facts vs Assumptions
- Facts come from files, tool output, user statements, or public evidence you actually read.
- Inferences are allowed only when clearly marked and relevant to packaging.
- If success cannot be externally checked, say so and keep the package conservative.
## Authorized Defaults
Usually safe after scope is locked:
- current thread context;
- current repository files and diffs;
- user-provided artifact directories;
- user-provided screenshots and logs;
- temporary local scratch files created for this task.
Require explicit approval:
- broad local conversation archives unrelated to the task;
- browser-session stores, cookies, local storage, or saved sessions;
- passwords, tokens, API keys, MFA material, or other credentials;
- private folders outside the agreed project scope.
## Workflow Reconstruction Output
Before writing the skill, produce:
- a success timeline;
- a failure-to-fix table;
- a state transition list;
- validation gates;
- unresolved assumptions.
If you cannot produce these, you do not yet understand the workflow well enough to package it.
## Reusable Invariants vs Local Accidentals
If you cite recurrence to support a reusable rule, state how many independent tasks showed the pattern, or say that their independence is unknown. Count continuations, forks, retries, and duplicate exports of the same task once. A retry that reproduces a failure strengthens the evidence for that failure within the task. A single completed workflow can still be packaged from its observed steps and outcomes.
Reusable:
- "choose the validated final artifact, not a raw intermediate";
- "check visible success evidence before claiming publication";
- "gate risky submission steps behind explicit confirmation".
Local-only:
- one temporary file path;
- one account name;
- one machine-specific directory;
- one-day UI wording;
- one-time manual preference not marked as durable.
## External Side Effects
For workflows that publish, upload, send, deploy, or mutate shared systems:
- capture the exact pre-submit gate;
- capture the observable success proof;
- capture the stop condition when evidence is missing;
- never reconstruct the workflow as "do the action and hope it worked".
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deployment | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| 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: 198k | Updated | Claude CodeCodex | Verified | Install |
| 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: 119k | Updated | Claude CodeCodex | Verified | Install |
| 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: 118k | Updated | Claude CodeCodex | Verified | Install |