Context Map Generation
SkillAI & modelsAnalyze the codebase to create a concise, LLM-optimized structured overview in .agent/map.md.
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the Context Map Generation skill
What this skill tells your AI
The instructions your AI receives, as published by diegosouzapw/awesome-omni-skill in skills/data-ai/agent-ops-context-map/SKILL.md and read by ahel’s review.
Purpose
Create a high-level, token-efficient overview of the system (.agent/map.md) to allow reasoning about the whole project without reading every file.
Confidence-Based Staleness Thresholds
Map freshness requirements scale with confidence level:
| Confidence | Max Age | Refresh Requirement |
|---|---|---|
| LOW | 24 hours | MANDATORY before implementation |
| NORMAL | 7 days | RECOMMENDED if significant changes |
| HIGH | 30 days | OPTIONAL |
Staleness Check
When invoked (or before low confidence work):
📍 CONTEXT MAP STALENESS CHECK
Map file: .agent/map.md
Last updated: {date} ({N} days ago)
Confidence level: {confidence}
Max age for confidence: {threshold}
{If stale:}
⚠️ Context map is STALE for {confidence} confidence work.
For LOW confidence, understanding the codebase is critical.
Refreshing map before proceeding...
{If fresh:}
✅ Context map is current ({N} days old, threshold: {threshold} days)
Low Confidence Refresh Requirements
For LOW confidence work:
- Check map age — if > 24 hours, refresh is MANDATORY
- Check for recent changes — git diff since last map update
- Partial refresh option — focus on affected areas if full refresh is expensive
🔄 LOW CONFIDENCE CONTEXT MAP REFRESH
Affected areas for {ISSUE-ID}:
- src/services/ (target of changes)
- src/models/ (dependencies)
- tests/services/ (test coverage)
Refresh options:
1. Full refresh (entire codebase)
2. Partial refresh (affected areas only)
3. Skip (NOT RECOMMENDED for low confidence)
Proceeding with partial refresh...
Procedure
- Scan the file structure (limit to 2-3 levels of depth).
- Identify key architectural elements:
- Critical configuration files (package.json, pyproject.toml, docker-compose, etc.)
- Entry points (main.py, index.js, App.tsx)
- Core modules and their responsibilities.
- Summarize architecture patterns and data flow.
- Write/Update
.agent/map.mdwith the following structure:- System Overview: One paragraph summary of purpose and stack.
- Key Components: List of major modules/folders and what they do.
- Patterns: Architectural decisions (e.g., MVC, Repository pattern, Event-driven).
- Key Files: Table of critical files and their specific role.
- Last Updated: Timestamp for staleness tracking.
- Constraint: Keep the file concise (target < 150 lines). It is a map, not a territory.
Map Header Template
Include this header in map.md for staleness tracking:
# Context Map
**Generated**: {YYYY-MM-DD HH:MM}
**Scope**: {full | partial: areas}
**Confidence**: {confidence level that triggered refresh}
---
Signals
- GitHub stars
- 57
- Forks
- 19
- Last commit
- Mar 2026
Advanced
- Catalog kind
- skill
- Gateway key
agent-ops-context-map- Source
- github.com/diegosouzapw/awesome-omni-skill