Does this skill work with existing documentation?

Yes, it complements existing documentation by providing high-level 'summaries' that point to specific files, helping the AI navigate your project more effectively.

Can agents write directly to the context folder?

No, the context folder is strictly read-only during active workflows. This ensures the reference material remains stable and predictable while the agent is performing tasks.

How does the context wiki stay updated?

The context is updated by a specialized Archival Skill once a workflow is completed, moving learnings from the active workspace into the permanent wiki.

Where is the context data stored?

All context data is stored locally within your project's .ai-dev-team/context/ directory, keeping your project knowledge inside your repository.

What is the main benefit of the Context Skill?

The primary benefit is token efficiency and situational awareness. It reduces the context needed for an agent to start working from 50,000+ tokens down to roughly 1,000-1,500 tokens.

AI Context Wiki Manager

Name: AI Context Wiki Manager
Author: marcel-Ngan

bymarcel-Ngan

•

生产力与工作流

Maintains a persistent, read-only wiki that provides high-level situational awareness to AI agents while reducing token consumption by up to 95%.

关于

The Context Skill establishes a structured wiki system within the .ai-dev-team directory to solve the problem of context drift and high token costs during long-running development tasks. By separating active workflow state from stable project knowledge, it allows AI agents to initialize with a condensed summary of the architecture, tech stack, and historical patterns instead of re-reading the entire codebase or conversation history. This ensures that every agent spawn starts with a consistent, high-level understanding of the project's source of truth without the overhead of processing thousands of redundant tokens.

主要功能

Solution-level summaries covering tech stacks and architecture patterns
Project-specific documentation for modular, on-demand context loading
Drastic token reduction, initializing context in approximately 1,000 tokens
1 GitHub stars
Historical story archiving to preserve learnings from past iterations
Read-only wiki architecture for stable agent situational awareness

使用场景

Maintaining a stable source of truth that remains unchanged during active coding tasks
Onboarding new AI agents to complex projects with minimal token overhead
Archiving historical project decisions and learned patterns for future AI reference

关于

主要功能

Solution-level summaries covering tech stacks and architecture patterns
Project-specific documentation for modular, on-demand context loading
Drastic token reduction, initializing context in approximately 1,000 tokens
1 GitHub stars
Historical story archiving to preserve learnings from past iterations
Read-only wiki architecture for stable agent situational awareness

使用场景

Maintaining a stable source of truth that remains unchanged during active coding tasks
Onboarding new AI agents to complex projects with minimal token overhead
Archiving historical project decisions and learned patterns for future AI reference