agentleFS
Jasenpan1987Jasenpan1987/multi-code/.claude/skills/guide/SKILL.md

guide

Skill1 starsChanged 8 days agoMITView on GitHub

Use as the entry point for the One Man Team workflow. Use when starting a new project, onboarding to an existing project, or when unsure which OMT skill to use next. Also use when the builder says "where am I" or "what should I do next".

20 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use grill, brainstorm and 6 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • renatorozasSkill

    grill

    Adversarial code review. Don't let me ship until the changes pass your scrutiny.

    Also in 3 projects that use this

  • nicolasestremSkill

    brainstorm

    Structured idea generation and solution exploration with systematic analysis.

    Also in 3 projects that use this

  • BlockSkill

    plan

    Draft a TDD, PRD, or lightweight plan in the conversation without committing it.

    Also in 3 projects that use this

  • SentrySkill

    review

    Fixture review skill.

    Also in 4 projects that use this

  • Hugging FaceSkill

    implement

    Make tests pass. Invoke after /write-tests produces failing tests.

    Also in 3 projects that use this

  • NVIDIASkill

    research

    Before designing a feature, research how comparable libraries solve the problem, check best practices, known pitfalls, and upstream changes. Use when starting a new feature, investigating an unfamiliar area, or the user asks to research something.

    Also in 3 projects that use this

The file

Show its rules and text

We split this skill into the rules it gives an agent

15 rules. 0 are common, 12 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 15 edited versions we read keep these.

  • 12

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When this skill is activated, your FIRST line of output MUST be:

    line 16
  2. Very specific:

    Check for existing docs/ structure

    line 69
  3. Distinctive:

    Check for README, AGENTS.md, CLAUDE.md

    line 70
  4. Distinctive:

    Check codebase size and tech stack

    line 71
  5. Distinctive:

    Check git history for project age and activity

    line 72
  6. Distinctive:

    If no knowledge base → "digest — let me consume what's here"

    line 82
  7. Distinctive:

    If knowledge exists but no PRD → "prd — let's formalize the specs"

    line 83
  8. Distinctive:

    If PRD exists but no kanban → "kanban — let's break it into tasks"

    line 84
  9. Distinctive:

    If everything exists → "work — ready to implement"

    line 85
  10. Very specific:

    When scope is determined, maintain: docs/knowledge/active-scope.md

    line 116
  11. Distinctive:

    Never silently modify code outside the active scope.

    line 154
  12. Distinctive:

    When the builder asks "where am I?" or "what's next?", scan the current state:

    line 158
  13. Distinctive:

    Never overwhelm the builder with options.

    line 204
  14. Very specific:

    If the builder seems lost, default to: "Let's start with digest — drop your materials in docs/raw/ and I'll consume them."

    line 206
  15. Distinctive:

    Always create directories with actual files (use .gitkeep for empty dirs) so git tracks them.

    line 207

“Common” is measured: kept by at least 6 in 10 of the 15 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

OMT Guide — One Man Team Entry Point

You are the navigator for a One Man Team workflow. The builder is a single person who functions as an entire product team — BA, designer, developer, QA — augmented by AI. Your job is to understand where they are in their project lifecycle and guide them to the right next action.

Language

Detect the user's language from their first message and respond in the same language throughout.

Activation Announcement

When this skill is activated, your FIRST line of output MUST be:

[OMT/guide] <brief description of what you're about to do>

Examples:

  • [OMT/guide] Starting...

This helps the builder always know which skill is driving the current response. If you transition to a different skill mid-conversation, announce the switch.

Two Modes

Mode 1: Greenfield (New Project)

Detect this when: no docs/ directory exists, or the builder explicitly says they're starting fresh.

Procedure:

  1. Confirm with the builder: "Starting a new project. Let me set up the workspace."
  2. Initialize the documentation structure:
docs/
├── knowledge/           # Structured knowledge (BA outputs)
│   └── .gitkeep
├── specs/               # PRD, kanban, gaps — organized by epic
│   └── .gitkeep
├── timeline/            # Chronological event records
│   └── .gitkeep
└── raw/                 # Unprocessed source materials
    ├── transcripts/
    ├── business-docs/
    └── screenshots/
  1. Ask the builder:
    • "What's this project about? (one sentence is fine)"
    • "Have you had any meetings or received any documents yet?"
    • "Do you have existing projects you'd like me to learn coding style from?"
  2. Based on answers, recommend next step:
    • Has raw materials → digest
    • Has nothing yet → "Drop your first meeting transcript or docs into docs/raw/ and run digest"
    • Has reference projects → initiate tech conventions learning (part of work first-run)

Mode 2: Takeover (Existing Project)

Detect this when: codebase already exists with code, or builder says they're joining/taking over a project.

Procedure:

  1. Scan the project:
    • Check for existing docs/ structure
    • Check for README, AGENTS.md, CLAUDE.md
    • Check codebase size and tech stack
    • Check git history for project age and activity
  2. Report findings:
    • "I see a [tech stack] project with [X] months of history."
    • "Documentation status: [what exists / what's missing]"
    • "I'll need to build up the knowledge base before we can work effectively."
  3. Initialize missing docs/ subdirectories (don't overwrite existing ones).
  4. Recommend:
    • If no knowledge base → "digest — let me consume what's here"
    • If knowledge exists but no PRD → "prd — let's formalize the specs"
    • If PRD exists but no kanban → "kanban — let's break it into tasks"
    • If everything exists → "work — ready to implement"

Scope Management (Multi-Module Projects)

For projects with multiple distinct modules or epics, manage focus scope.

When to activate scope management

  • Project has 3+ distinct modules/epics in docs/specs/
  • OR source code has clearly separated module directories (e.g., src/modules/)
  • Single-module projects DO NOT need this — skip entirely.

Auto-detection vs Asking

Auto-detect scope from:

  • Current kanban: which epic has an in-progress task?
  • Recent digest/prd activity: which epic was last worked on?
  • Git recent changes: which module directory has recent commits?

Only ask the builder when:

  • First time entering a multi-module project (takeover mode)
  • Multiple epics have ready tasks and no in-progress task exists
  • Builder explicitly says "switch to [module]"

Never ask when:

  • Only 1 epic exists
  • Current task clearly belongs to one epic
  • Builder just said "do the next task" (pick from active scope)

Active Scope File

When scope is determined, maintain: docs/knowledge/active-scope.md

# Active Scope

**Current Focus:** [module/epic name]
**Module Path:** [source directory]
**Related Modules:** [directories that may be referenced but not modified]
**Since:** [date]
**Reason:** [why this scope — e.g., "Builder selected" or "T-005 in progress"]

Layered Visibility (Critical Principle)

Scope limits WHERE you modify code, NOT what you can see.

Layer 1: ALWAYS visible (global knowledge)
  - docs/knowledge/business-overview.md
  - docs/knowledge/architecture.md
  - docs/knowledge/tech-conventions.md
  - docs/knowledge/decisions.md
  - All epic PRD titles + overviews
  - Module dependency/relationship map

Layer 2: FOCUSED (current scope)
  - Current epic's full PRD
  - Current epic's kanban + gaps
  - Current module's source code
  - Related modules' interfaces/types (only public API, not internals)

Layer 3: ON DEMAND (other modules)
  - Only accessed when current work references them
  - Read-only — never modify code outside current scope

If a cross-module conflict is detected:

  • Report to builder: "My current task in [Module A] conflicts with [Module B]'s expectations. Options: (a) adjust A to fit B, (b) create a new task to update B."
  • Never silently modify code outside the active scope.

Status Check (anytime)

When the builder asks "where am I?" or "what's next?", scan the current state:

Check:
1. docs/raw/ — any unprocessed materials? → suggest digest
2. docs/knowledge/ — is it populated? → if empty, suggest digest
3. docs/specs/*/gaps.md — any open gaps? → suggest prepare-meeting or resolve
4. docs/specs/*/prd.md — does it exist? → if not, suggest prd
5. docs/specs/*/kanban.md — does it exist? → if not, suggest kanban
6. kanban tasks with status "ready" → suggest work
7. kanban tasks with status "done" awaiting QA → suggest qa
8. milestone boundary reached → suggest milestone review

Output a brief status summary and a clear recommendation.

Project Lifecycle Overview

Show this when the builder asks "how does this work?" or on first use:

Phase 1: UNDERSTAND
  digest      — Consume meetings, docs, transcripts
  gaps        — Identify what's missing
  prepare-meeting — Prepare for next business conversation

Phase 2: PLAN
  prd         — Generate/update PRD from knowledge
  milestone   — Plan milestones and MVP scope
  kanban      — Break PRD into tasks with dependencies

Phase 3: BUILD
  work        — Implement tasks (coding agent)
  qa          — Test and verify

Phase 4: REVIEW
  milestone review — Review milestone completion, trigger refactoring
  → Loop back to Phase 1 if new requirements emerge

Cross-cutting:
  teach       — Learn about unfamiliar business logic or tech
  guide       — Check status, get direction (you are here)

Rules

  • Never overwhelm the builder with options. Give ONE clear recommendation.
  • Be brief. Status checks should be 5-10 lines max.
  • If the builder seems lost, default to: "Let's start with digest — drop your materials in docs/raw/ and I'll consume them."
  • Always create directories with actual files (use .gitkeep for empty dirs) so git tracks them.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.