agentsclimarketplace

Business tracker

Skill k1r3zz/kira-claude-code/skills/business-tracker/skills/business-tracker/business-tracker

Auto-extract business logic from any codebase into living documentation. Scans code to discover business rules, tracks changes across sessions, and maintains always-up-to-date rule docs organized by module. Supports Flutter, Node.js, Go, Python, Java, .NET, Rust, Ruby, PHP, iOS. Triggers: 'biz scan', 'biz-scan', 'scan business logic', 'biz update', 'record business change', 'biz query', 'document business logic'. Use this skill when the user explicitly wants to scan, extract, document, or track business rules in a codebase. Do NOT trigger for general code questions, debugging, or feature development — only when the user's intent is specifically about business logic documentation.From its SKILL.md

Install
npx -y skills add k1r3zz/kira-claude-code --skill business-tracker

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.
  • 1 stars1 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

5.5 KB, ~1.1k tokens by cl100k_base, as published. Nobody here has run it

Business Tracker

Auto-extract, continuously track, and maintain living documentation of business logic from any codebase.

Works with any tech stack: Flutter, Node.js, Go, Python, Java/Kotlin, .NET, Rust, Ruby, PHP, iOS (Swift/Objective-C).

How It Works

Trigger via /business-tracker + natural language. The skill matches user intent to the right command:

User saysActionCommand file
"scan business logic", "biz scan", "scan src/services/"Scancommands/biz-scan.md
"record business change", "biz update", "update business docs"Track changecommands/biz-update.md
"what's the payment rule?", "biz query", "query business rules"Querycommands/biz-query.md
"关联 openspec", "link openspec", "biz link"Link to OpenSpeccommands/biz-link-openspec.md

Intent matching:

  • Contains "scan" → run scan (if a path follows, scope to that path)
  • Contains "record", "update", "change" → track a change
  • Contains "query", "what rule", "how does X work" → query rules
  • Contains "关联", "link" + "openspec" → link biz docs to openspec
  • Ambiguous → ask the user

On receiving a command: read the corresponding file under this skill's commands/ directory and follow its instructions. Scan strategy and detection rules live in references/ — load them on demand, not upfront.

Output Structure

biz_scan/
├── index.md                    # Master index (project overview + module list)
├── modules/
│   ├── <module_name>.md        # Detailed rules per business module
│   └── ...
├── changelog.md                # Business change history (append-only)
└── .scan-cache.json            # Scan cache (git hash + progress, do not edit)

Progressive Disclosure (@ references)

  • index.md contains only the project overview + one-line summary per module + @biz_scan/modules/xxx.md reference
  • Queries load index.md first, then only the relevant module file(s)
  • Each module file is self-contained and small — context-friendly

Commit everything to Git for team sharing.

Core Principles

  1. Cover every rule — Every flow, every branch, every constraint gets documented. Not a high-level summary.
  2. Write product docs, not code manuals — Organize by process steps and natural language, not by code structure.
  3. Flows first, rules second — Describe end-to-end flows (numbered steps) before listing standalone rules.
  4. Don't repeat what code already tells you — Skip class names, field lists, method signatures, API params. Only keep entry-point file paths.
  5. Auto-split large files — When a module exceeds 150 lines, auto-split into sub-files by business topic.
  6. Record the why, not just the what — If the reason is unknown, mark it "reason TBD".
  7. Mark uncertainty — Uncertain rules get a [speculative] tag. Encourage human confirmation.
  8. Transparent coverage — After scanning, report exactly what was scanned, skipped, and remaining.
  9. Human-AI collaboration — Auto-generate the first draft; encourage manual edits and additions.
  10. Match user's language — All generated documentation (index.md, module files, changelog, coverage report, progress output) MUST be written in the same language the user used to invoke the command. If the user writes in Chinese, all docs are in Chinese. If in English, all docs are in English. Code identifiers (class names, file paths, variable names) always stay in their original form.

Supported Stacks

The skill auto-detects project type and loads only the relevant scan rules:

  • Mobile: Flutter/Dart, iOS (Swift/Objective-C)
  • Frontend: React/Vue/Angular (TypeScript/JavaScript)
  • Backend: Node.js, Go, Python (Django/FastAPI), Java/Kotlin (Spring), .NET, Rust, Ruby (Rails), PHP (Laravel)
  • Monorepo: Detects each subdirectory independently

Reference Files

FileContentWhen to load
references/scan-rules/common.mdProject type detection, business signal patterns, batching strategy, cache formatbiz-scan and biz-update
references/scan-rules/<stack>.mdStack-specific file priority rules (e.g., flutter.md, node.md)After detecting project type
references/index-template.mdTemplate for index.mdFirst-time doc generation
references/module-template.mdTemplate for module files (single-file format)Creating new module files
references/module-index-template.mdTemplate for split module indexWhen module exceeds 150 lines and needs splitting
references/module-subfile-template.mdTemplate for split module sub-topic filesCreating sub-topic files under split modules
references/changelog-template.mdTemplate for changelog entriesbiz-update

What ships with it: 22 files

57.9 KB alongside SKILL.md

Keep looking

Skills are one crate of 326,758. 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.