document-manager
How to Install
Claude Code:
git clone --depth 1 https://github.com/BFLabsAI/algoritmo-milionario.git && cp algoritmo-milionario/skills/document-manager ~/.claude/skills/document-manager -r---
name: document-manager
description: >
Full-spectrum document and planning manager. Use for organizing files and folders (clean up Downloads, find duplicates, restructure directories), writing technical documentation with the Diátaxis framework (tutorials, how-to guides, reference, explanation), creating and executing implementation plans (phased sprints, atomic tasks, TDD-style step-by-step plans), designing architecture decisions (ADR creation, trade-off analysis, technology comparison), auditing and improving CLAUDE.md files across repositories, and CLAUDE.md maintenance or project memory optimization. Also triggers on: "plan harder", implement a plan, write a plan, create implementation plan, organize my files, document this, architecture decision, ADR.
tools: Read, Glob, Grep, Bash, Edit
---
# Document Manager
Unified skill for the full lifecycle of documents, plans, and knowledge artifacts: file organization, technical writing, implementation planning, architecture decisions, and CLAUDE.md quality management.
---
## Domain 1: File Organization
Intelligently organizes files and folders by understanding context, finding duplicates, suggesting structures, and automating cleanup.
**Use when:** Downloads folder is chaotic, files are scattered, duplicates waste space, folder structure needs restructuring, starting/archiving a project.
### Workflow
1. **Scope check** — Ask: which directory, main problem, files to avoid, aggressiveness level.
2. **Analyze** — Survey file types, sizes, counts, and date ranges.
3. **Propose plan** — Present proposed folder structure and move list. Await approval.
4. **Execute** — Create folders, move/rename files with logging. Always confirm before deleting.
5. **Summary** — Report changes, provide maintenance schedule.
**Rules:**
- Always confirm before deleting anything.
- Preserve original modification dates.
- Handle filename conflicts gracefully.
- Stop and ask on unexpected situations.
For detailed bash commands, organization patterns, and naming conventions, see [references/file-organization.md](references/file-organization.md).
---
## Domain 2: Technical Documentation (Diátaxis)
Expert technical writer guided by the Diátaxis framework. Creates tutorials, how-to guides, reference docs, and explanations.
**Use when:** Writing software docs, creating API references, drafting tutorials, or structuring explanation content.
### Diátaxis Document Types
| Type | Purpose | Analogy |
|------|---------|---------|
| Tutorial | Learning-oriented, practical steps for newcomers | A lesson |
| How-to Guide | Problem-oriented, steps to solve a specific problem | A recipe |
| Reference | Information-oriented, technical description of machinery | A dictionary |
| Explanation | Understanding-oriented, clarifying a topic | A discussion |
### Workflow
1. **Clarify** — Determine document type, target audience, user's goal, and scope.
2. **Propose structure** — Outline with section descriptions. Await approval.
3. **Generate** — Write full Markdown content adhering to: Clarity, Accuracy, User-Centricity, Consistency.
**Rules:**
- Use context from provided Markdown files for tone/style, but do not copy content unless asked.
- No external sources unless explicitly provided.
---
## Domain 3: Implementation Planning
Three complementary planning modes — choose based on the request:
### Mode A: Plan Harder (Phased Sprint Planning)
**Trigger:** User says "plan harder"
Creates detailed phased plans with sprints and atomic tasks. Process:
1. Research codebase — architecture, patterns, dependencies.
2. Clarify requirements — ask up to 10 targeted questions.
3. Create plan — Sprints → Tasks → Gotchas → Review.
4. Save as `[name]-plan.md`, then review with a subagent.
Each sprint must produce a demoable, runnable, testable increment. Each task must be atomic, committable, and independently testable.
See [references/plan-harder-template.md](references/plan-harder-template.md) for the full sprint/task template.
### Mode B: Writing Plans (TDD Step-by-Step)
**Trigger:** Have a spec or requirements for a multi-step task, before touching code.
Announces: "I'm using the writing-plans skill to create the implementation plan."
- Map file structure before decomposing tasks.
- Each step is one action (2–5 min): write failing test → run it → implement → verify → commit.
- No placeholders. Every step has actual code, exact paths, exact commands with expected output.
- Save to `docs/superpowers/plans/YYYY-MM-DD-.md` (or user-specified location).
- After saving, offer two execution options: Subagent-Driven (recommended) or Inline Execution.
See [references/writing-plans-guide.md](references/writing-plans-guide.md) for plan header template, task structure, and self-review checklist.
### Mode C: Create Implementation Plan (AI-Optimized Format)
**Trigger:** Create implementation plan for feature/refactor/upgrade/architecture/infrastructure.
Produces machine-readable, deterministic plans for autonomous AI or human execution.
- Plans saved to `/plan/` with naming `[purpose]-[component]-[version].md`.
- Purpose prefixes: `upgrade|refactor|feature|data|infrastructure|process|architecture|design`.
- Front matter includes: goal, version, date, owner, status (with badge), tags.
- Structured with phases, TASK-NNN identifiers, completion criteria, and parallel-safe task grouping.
See [references/implementation-plan-template.md](references/implementation-plan-template.md) for the full mandatory template.
### Executing Plans
**Trigger:** Have a written implementation plan to execute.
Announces: "I'm using the executing-plans skill to implement this plan."
Process:
1. Load and review plan critically — raise concerns before starting.
2. Execute each task: mark in_progress → follow steps exactly → run verifications → mark complete.
3. **Stop immediately** when: blocker hit, instruction unclear, verification fails. Ask for clarification — never guess.
4. On completion, transition to finishing-a-development-branch workflow.
**Rule:** Never start implementation on main/master branch without explicit user consent.
---
## Domain 4: Architecture Decisions (ADR)
Create or evaluate Architecture Decision Records. Use when choosing technologies, documenting design decisions, reviewing system proposals, or designing new components.
**Use when:** "Should we use X or Y?", "Review this design", "Design the notification system", "Document why we chose X".
### ADR Output Format
```markdown
# ADR-[number]: [Title]
**Status:** Proposed | Accepted | Deprecated | Superseded
**Date:** [Date]
**Deciders:** [Who needs to sign off]
## Context
## Decision
## Options Considered
(table: Complexity, Cost, Scalability, Team familiarity | Pros | Cons)
## Trade-off Analysis
## Consequences
## Action Items
```
**Tips:**
- State constraints upfront (timeline, scale requirements).
- Name all options explicitly, even if leaning one way.
- Include non-functional requirements: latency, cost, team expertise, maintenance burden.
---
## Domain 5: CLAUDE.md Quality Management
Audit, evaluate, and improve CLAUDE.md files across a repository to ensure Claude Code has optimal project context.
**Trigger phrases:** Check/audit/update/improve/fix CLAUDE.md, CLAUDE.md maintenance, project memory optimization.
### Workflow
1. **Discover** — Find all CLAUDE.md files: `find . -name "CLAUDE.md" -o -name ".claude.md" -o -name ".claude.local.md"`.
2. **Assess** — Score each file against 6 criteria (see [references/claude-md-quality-criteria.md](references/claude-md-quality-criteria.md)).
3. **Report** — Output quality report BEFORE making any changes. Grades: A(90-100), B(70-89), C(50-69), D(30-49), F(0-29).
4. **Confirm** — Ask for user approval before updating.
5. **Update** — Apply targeted additions using Edit tool. Preserve structure.
### Quality Criteria Summary
| Criterion | Weight |
|-----------|--------|
| Commands/Workflows documented | 20pts |
| Architecture clarity | 20pts |
| Non-obvious patterns/gotchas | 15pts |
| Conciseness | 15pts |
| Currency (reflects current state) | 15pts |
| Actionability (copy-paste ready) | 15pts |
### Key Principles
- **What TO add:** Commands discovered, gotchas/non-obvious patterns, package relationships, testing approaches that work, config quirks.
- **What NOT to add:** Obvious code info, generic best practices, one-off fixes, verbose explanations.
- Use `#` shortcut during sessions for Claude to auto-incorporate learnings.
- Personal/local preferences go in `.claude.local.md` (gitignored).
- User-wide defaults go in `~/.claude/CLAUDE.md`.
See [references/claude-md-templates.md](references/claude-md-templates.md) and [references/claude-md-update-guidelines.md](references/claude-md-update-guidelines.md) for templates and detailed update guidance.
---
## Rules and Invariants
- **File operations:** Always confirm before deleting. Log all moves. Stop on unexpected situations.
- **Plans:** Never start implementation on main/master without explicit user consent. Stop when blocked — never guess.
- **Documentation:** Determine document type before writing. Propose structure before generating content.
- **Plans — no placeholders:** TDD-style plans must contain actual code, exact paths, exact commands. Never write "TBD", "TODO", "implement later", "similar to Task N".
- **CLAUDE.md:** Always output quality report before making changes. Get user approval before any edits.
- **Conflict resolution:** File cleanup ("reduces clutter") applies to personal workspace files. Writing-plans code file advice ("don't unilaterally restructure") applies to codebase source files — no conflict.
## When to Use
| Request | Mode |
|---------|------|
| "Organize my Downloads / find duplicates / clean up folders" | Domain 1: File Organization |
| "Write documentation / tutorial / how-to / reference" | Domain 2: Diátaxis Documentation |
| "Plan harder" | Domain 3A: Phased Sprint Planning |
| "Write a plan for this spec" | Domain 3B: TDD Step-by-Step Plans |
| "Create implementation plan for feature/refactor/upgrade" | Domain 3C: AI-Optimized Plan |
| "Execute this plan / implement the plan" | Domain 3: Executing Plans |
| "Architecture decision / ADR / should we use X or Y" | Domain 4: Architecture Decisions |
| "Check/audit/improve CLAUDE.md / project memory" | Domain 5: CLAUDE.md Quality |
Details
| Category | Coding → code-quality |
| Source | BFLabsAI/algoritmo-milionario |
| SKILL.md | View on GitHub → |
| Repo Stars | N/A |
| Est. per Skill | N/A (shared across 30 skills from this repo) |
| Difficulty | Advanced |
| Risk Level | Safe |
Related Skills
speed
Speed Reader Launch the RSVP speed reader to display text one word at a time with Spritz-style ORP (
defuddle
Defuddle Use Defuddle CLI to extract clean readable content from web pages. Prefer over WebFetch for
xlsx-official
Requirements for Outputs All Excel files Zero Formula Errors Every Excel model MUST be delivered wit
speed
Speed Reader Launch the RSVP speed reader to display text one word at a time with Spritz-style ORP (
Works Well With
Skills from the same repository — often designed to work together
github-ops
--- name: github-ops description: Unified GitHub operations skill covering git commit workflows (con
prompt-engineer
--- name: prompt-engineer description: >- Full-stack prompt engineering skill — covers every stage o
find-skills
--- name: find-skills description: Helps users discover and install agent skills when they ask quest