From contextforge
Context-aware coding intelligence with automatic codebase indexing, semantic RAG retrieval, and persistent cross-session memory
How this skill is triggered — by the user, by Claude, or both
Slash command
/contextforge:contextforgeThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
You have access to injected codebase context from ContextForge. Use this context to provide more accurate, project-aware responses.
You have access to injected codebase context from ContextForge. Use this context to provide more accurate, project-aware responses.
ContextForge injects several types of context automatically. When you see these sections, prioritize them over your general training knowledge:
A compact summary of the codebase structure including:
How to use: Reference this to understand the project's overall structure before suggesting changes. Ensure your suggestions align with the existing architecture.
Semantically retrieved code chunks that are relevant to the current user prompt. Each chunk includes:
How to use: These are the most relevant pieces of existing code to the user's question. Reference specific files and line numbers. Build on existing patterns rather than inventing new ones. If the retrieved context shows a particular coding style, follow it.
Coding patterns that have been observed multiple times across the project:
How to use: Always follow these conventions when writing new code. They represent the team's established patterns. For example, if the project uses UUID keys, use UUIDs for new entities. If it uses the repository pattern, create repositories for new data access.
Files that have been changed in recent sessions, indicating areas of active development.
How to use: Be aware of recent changes to avoid conflicts and to understand current development focus.
npx claudepluginhub kaakati/context-forge --plugin contextforgeBuilds a mental model of unfamiliar codebases by searching, reading dependencies, and checking git history. Use when onboarding or before editing unknown code.
Establishes that Cortex context must be consulted before answering code questions, and dispatches to the right Cortex skill for research, impact, and review tasks.
Navigates large codebases via MCP server: symbol search, reference/caller lookups, commit history, blame, diffs, and code outlines without reading source files.