ai_tooling

Generic, reusable tooling and knowledge bases for working with AI coding agents: CLIs for navigating and documenting codebases, a transverse level-of-detail practice, and guideline knowledge bases an agent can consult mid-task.

Status: the working toolkit of the IntentGraph project, open source (MIT for code, CC BY 4.0 for the knowledge bases). The three CLIs are small standalone Python projects with their own tests; the guidelines are living documents. Interfaces and content may change without notice.

Getting It

This repository is designed to sit beside the sibling intent_graph repository: relative links from here assume a checkout layout of ../intent_graph. Clone both into the same folder (the second one is optional, the CLIs below do not need it):

git clone https://github.com/intent-graph/ai_tooling.git
git clone https://github.com/intent-graph/intent_graph.git

Layout

• documentation_tools/ — md_structure_sync (TOC/line-range sync CLI), md_structure_viz (doc-tree visualization CLI), codemap (structural code-graph CLI for Python), and DOCUMENTATION_STANDARD.md (the doc format these tools maintain).
• dev_guidelines/ — software_development_guidelines (stack-neutral principles: what a team writes, adopts or rents, how it assures, changes, operates and evolves a system, held as one IntentGraph), python_guidelines (a Python knowledge base, LOD-organized), cli_guidelines (CLI interface-design consensus: clig.dev/12-Factor/POSIX/GNU), web_architecture_guidelines (architecture of a JavaScript + Python + Postgres web application: who does what across tiers, held as two IntentGraphs) and javascript_guidelines (solid, maintainable, secure JavaScript at the language level, from audited authoritative sources, held as two IntentGraphs).
• diagrams_for_humans/ — reference for generating yEd .graphml diagrams.
• level_of_detail/ — the level-of-detail practice: an agent-facing core plus satellites, and the authoring record behind them.
• howto/ — short how-tos (exposing a Python tool system-wide, writing a structured subagent).
• global_skills/ — Claude Code skills meant for every project, installed user-wide by symlink (subagent-orchestration: brief sub-agents by pointer, fixed I/O, three-level returns; session-handoff: hand a task to a new session, intent first, pointers to disk).
• ai_evolution/ — notes on AI evolution and method (in French).
• .claude/ — agents (doc-manifest-agent, diagram-agent, python-kb-agent) and skills (md-doc-tools, codemap) that wrap the CLIs above for agent use.

Installing the CLIs

Each CLI is a standalone Python project (Python 3.11+) installed system-wide with uv, from the root of this repository:

uv tool install -e documentation_tools/codemap
uv tool install -e documentation_tools/md_structure_sync
uv tool install -e documentation_tools/md_structure_viz

This puts codemap, md-structure-sync, md-doc-tree-simple, md-doc-tree-rich, and md-doc-tree-full on PATH, callable from any directory. Check each tool's own README for test setup and alternative install methods. Check the install from the repository root:

md-doc-tree-rich howto                                   # Markdown files with line counts and titles
codemap overview documentation_tools/md_structure_sync   # structural map of a Python project (YAML)

Contributing

Issues are welcome. A pull request gives its content entirely to the owner: read CONTRIBUTING.md first.

License

• CC BY 4.0 — level_of_detail/, ai_evolution/, dev_guidelines/, howto/, documentation_tools/DOCUMENTATION_STANDARD.md (intellectual/creative work).
• MIT — everything else: documentation_tools/ code, diagrams_for_humans/, .claude/, global_skills/.
• Code files inside the CC BY folders (*.py, *.sh) are MIT.

Copyright 2026 Rémi Cura. Texts: LICENSE (MIT) and LICENSE-CC-BY-4.0.txt; each CC BY folder also carries a short LICENSE.md.