Skip to content

About

Analyze a codebase or documented project plan and generate a structured development backlog with prioritized tasks, coding conventions, and multi-agent configuration files.

Topics

Resources

Stars

1 star

Watchers

0 watching

Forks

Latest commit

 

History

1 Commit

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 

Repository files navigation

project-analysis

Analyze a codebase or documented project plan and generate a structured development backlog with prioritized tasks, coding conventions, and multi-agent configuration files.

What It Does

This skill runs a 4-phase analysis on your project and produces:

  • Task backlog (tasks/BACKLOG.md) with prioritized, dependency-ordered waves
  • Individual task files (tasks/*.md) with acceptance criteria, test requirements, and references
  • Agent workflow guide (tasks/README.md) with PR checklists and branch naming conventions
  • Coding conventions (.ai/rules/) as a single source of truth for all AI tools
  • Agent entry points for Claude Code, GitHub Copilot, JetBrains Junie, Cursor, and Aider
  • Quality baseline with coverage, lint, complexity, and dependency health metrics

It works on existing codebases and documentation-first/greenfield projects alike.

Install

OpenCode (global)

mkdir -p ~/.config/opencode/skills/project-analysis
cp -r . ~/.config/opencode/skills/project-analysis/

OpenCode (project-local)

mkdir -p .opencode/skills/project-analysis
cp -r . .opencode/skills/project-analysis/

Claude Code (global)

mkdir -p ~/.claude/skills/project-analysis
cp -r . ~/.claude/skills/project-analysis/

Claude Code (project-local)

mkdir -p .claude/skills/project-analysis
cp -r . .claude/skills/project-analysis/

Generic agents path

mkdir -p .agents/skills/project-analysis
cp -r . .agents/skills/project-analysis/

The directory name must be project-analysis (matching the skill name).

Usage

Trigger the skill with any of these phrases:

  • "analyze this project"
  • "create a task backlog"
  • "onboard this codebase"
  • "what needs fixing"
  • "set up coding standards"
  • "generate tasks"
  • "scaffold from docs"
  • "build this from the spec"

How It Works

Phase 1 -- Deep Analysis

Reads all source files, configs, tests, and documentation. Extracts coding conventions, categorizes findings by severity (Critical/High/Medium/Low), triages TODO/FIXME/HACK markers, measures quality baselines, and detects existing AI tool configurations.

Phase 2 -- Prioritize and Group

Assigns priority and size estimates to each finding, organizes them into dependency-ordered waves (A, B, C, ...), and assigns unique task IDs.

Checkpoint -- User Validation

Before generating files, the skill pauses and presents:

  1. Summary of findings by priority
  2. Proposed waves with themes and task counts
  3. Key architectural observations
  4. Quality baseline snapshot
  5. AI tool selection (see below)

You confirm or adjust priorities, scope, and tool selection before proceeding.

Phase 3 -- Generate Output

Generates all task files, the backlog, workflow guide, coding conventions, and agent entry points for the tools you selected.

Phase 4 -- Quality Check

Validates all generated output against a checklist: cross-references, formatting, completeness, and consistency.

AI Tool Selection

The skill auto-detects which AI tools are already configured in your project by checking for known config files (.claude/, .github/copilot-instructions.md, .junie/, .cursor/rules/, .aider.conf.yml). Detected tools are pre-selected at the Checkpoint.

You then confirm or adjust the selection. Only pointer files for confirmed tools are generated.

Always generated (regardless of selection):

  • AGENTS.md -- Universal agent entry point
  • .ai/rules/ -- Canonical coding conventions (single source of truth)
  • .editorconfig -- Editor formatting (if missing)

Conditionally generated (based on your selection):

Tool Generated files
Claude Code .claude/CLAUDE.md
GitHub Copilot .github/copilot-instructions.md, .github/instructions/, .vscode/settings.json
JetBrains Junie .junie/guidelines.md
Cursor .cursor/rules/project-conventions.mdc
Aider .aider.conf.yml

All pointer files reference .ai/rules/ -- no rule duplication across tools.

Output Structure

AGENTS.md
.ai/rules/
  project.md
  code-style.md
  testing.md
  security.md
  quality.md
.editorconfig
tasks/
  BACKLOG.md
  README.md
  A1-fix-auth-bypass.md
  B1-add-input-validation.md
  ...
.claude/CLAUDE.md                    # if selected
.github/copilot-instructions.md     # if selected
.github/instructions/               # if selected
.vscode/settings.json               # if selected
.junie/guidelines.md                 # if selected
.cursor/rules/                       # if selected
.aider.conf.yml                      # if selected

Re-Running

Re-run the skill after major refactors, wave completions, dependency overhauls, or when the backlog is >30% outdated. The skill preserves completed and in-progress tasks and only updates pending items.

File Inventory

File Purpose
SKILL.md Main entry point and orchestrator
references/01-read-project.md What to read and how to map architecture
references/02-extract-conventions.md Convention extraction instructions
references/03-identify-findings.md Issue categorization (Critical/High/Medium/Low)
references/04-triage-todos.md TODO/FIXME/HACK triage process
references/05-measure-baseline.md Quality baseline measurement
references/06-prioritize-and-group.md Priority, sizing, wave grouping, IDs
references/07-task-format.md Per-task file template
references/08-backlog-format.md BACKLOG.md template
references/09-workflow-format.md Agent workflow guide and PR checklist
references/10-rules-format.md .ai/rules/ file templates
references/11-agent-pointers.md Multi-agent entry points and pointer templates
references/12-quality-checklist.md Verification checklists for all output
references/additional-rules.md Optional domain-specific rule templates

License

MIT

About

Analyze a codebase or documented project plan and generate a structured development backlog with prioritized tasks, coding conventions, and multi-agent configuration files.

Topics

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors