Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsCommunityBlog
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

Back to skills

Project Cartography

ASecurity

Use when starting a project expected to outgrow one session, when joining a large unfamiliar codebase, or when a session begins in a project that keeps CODEMAP.md/PROJECT_STATE.md. Maintains a living three-file map (what is where, what is happening, why decisions were made) so each session starts from ~300 lines instead of re-reading the codebase.

2 stars
0 votes
0 copies
0 views
Added 9/19/2026
ai-agentsrustgitapi

Works with

api

Security Analysis

A100/100

Scanned 9/19/2026

Install to Claude Code

$npx -y skills add Mixard/fable-pack --skill project-cartography --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Project Cartography?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Project Cartography
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/mixard-project-cartography/badge)](https://www.skillsdirectory.com/skills/mixard-project-cartography)

More formats (shields.io, HTML) on the badges page.

Download Zip
Files
SKILL.md
---
name: project-cartography
description: Use when starting a project expected to outgrow one session, when joining a large unfamiliar codebase, or when a session begins in a project that keeps CODEMAP.md/PROJECT_STATE.md. Maintains a living three-file map (what is where, what is happening, why decisions were made) so each session starts from ~300 lines instead of re-reading the codebase.
---

# Project cartography

A frontier model's binding constraint on large projects is context, not intelligence.
Without a map, every session pays a re-reading tax; with a stale map, it pays twice.
This skill maintains three small files that convert "re-read the project" into
"read 300 lines".

**Prime directive: the map serves the model, it does not command it.** These files
transfer context between sessions - they are not process artifacts. If a file stops
earning its tokens, shrink it. Never let map upkeep displace judgment: the model
decides what is substantive, what to record, and what to skip.

## The three files (project root)

**CODEMAP.md** - what is where. One line per directory/module: purpose + the key
entry point. Not a file listing - a reading guide. Target under 100 lines even for
large repos; if a directory needs three lines, it needs a refactor note instead.

```markdown
# CODEMAP
- src/api/ - HTTP layer; routes registered in src/api/router.ts
- src/core/pricing/ - all price math; entry: calculate.ts, pure functions only
- src/jobs/ - cron workers; each file self-registers, see jobs/index.ts
- migrations/ - append-only; never edit applied ones
Danger zones: src/legacy/ (do not touch without reading DECISIONS.md#legacy-freeze)
```

**PROJECT_STATE.md** - what is happening. Done / in progress / next, with dates.
The handoff note a session writes to its successor. Prune ruthlessly: done items
older than a few weeks collapse into one line.

```markdown
# PROJECT STATE (updated 2026-07-21)
## In progress
- payments: Stripe webhook retries - src/jobs/stripe_retry.ts exists, untested
## Next
- rate limiting on /api/quote (decided, see DECISIONS.md#rate-limits)
## Done (recent)
- 2026-07-19: auth moved to sessions, JWT code deleted
```

**DECISIONS.md** - why. Append-only log of choices that future work must not
relitigate: one heading per decision, 3-6 lines each (context, choice, rejected
alternatives, consequences). Record decisions, not discussions.

## When to scaffold

Scaffold at project start only when the project will plausibly outgrow one session -
a script or small tool does not need cartography, and empty ceremony files are worse
than none. In an existing large codebase, generate CODEMAP.md from exploration once,
then maintain.

On scaffold, add one standing rule to the project's CLAUDE.md:

```markdown
## Project map
CODEMAP.md, PROJECT_STATE.md, DECISIONS.md at root are the session handoff. Read them
before exploring. After substantive changes (new module, moved responsibility, finished
or abandoned task, architectural choice), update the affected file - substantive is
your judgment call. Renames, small fixes, and refactors that change no responsibility
do not warrant an update.
```

## Update rules

- Update at natural checkpoints (task finished, direction changed), not per-edit.
- CODEMAP changes only when the answer to "where does X live" changes.
- STATE is rewritten freely; CODEMAP is edited surgically; DECISIONS is append-only.
- Session start in a mapped project: read the three files INSTEAD of broad exploration,
  then verify only the parts you will touch - the map can be stale; trust but verify.
- Planning a large feature (writing-plans skill): the plan's phases land in
  PROJECT_STATE.md "Next" so expansion survives the session.

## Hard rules

- Never block or delay actual work to polish the map - a 2-minute update at a
  checkpoint, not a bookkeeping session.
- Never duplicate what git already records (history, authorship, diffs) or what
  CLAUDE.md already says (conventions, commands).
- Combined size of the three files stays under ~400 lines; beyond that the map has
  become a second codebase - compress it.

Attribution

MixardMixard
View sourceMore from Mixard →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accuracy. Supports intensity levels: lite, full (default), ultra, wenyan-lite, wenyan-full, wenyan-ultra. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman. Also auto-triggers when token efficiency is requested.

1023331 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

686011 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3331 votes

catchup

Recovers prior coding-agent session context by running `catchup <agent> --since-compact`, which extracts a clean summary of a previous Codex, Claude Code, Antigravity, OpenCode, or Pi Agent session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", or asks to recover/summarize a previous session before continuing. Do NOT use for the current conversation, git history, or any non-agent log.

611 votes

math-skill

A comprehensive mathematical reasoning skill for AI assistants — handles arithmetic to research-level problems with rigorous step-by-step reasoning, systematic verification, and transparent uncertainty handling

381 votes
View all in ai-agents →