Skip to content

Features

Overview

graph TD FC[".fcontext/"] FC --> MA["🤖 Multi-Agent<br/>Support"] FC --> DI["📄 Document<br/>Indexing"] FC --> DC["🧠 Dynamic Context<br/>Building"] FC --> EP["📦 Experience<br/>Packs"] FC --> RM["📋 Requirements<br/>Management"] FC --> WM["🗺️ Workspace<br/>Map"] FC --> LS["📝 Living<br/>Summary"] FC --> OS["🔒 Offline &<br/>Secure"]

Multi-Agent Support

fcontext generates instructions in each agent's native format. Enable any agent with one command:

fcontext enable copilot    # .github/instructions/*.instructions.md
fcontext enable claude     # .claude/rules/*.md
fcontext enable cursor     # .cursor/rules/*.md
fcontext enable trae       # .trae/rules/*.md
fcontext enable opencode   # uses Claude format
fcontext enable openclaw   # skills/ only

All agents share the same .fcontext/ data — switch freely without losing context.

Document Indexing

Convert binary files to Markdown so any AI agent can read them:

fcontext index specs/requirements.pdf
fcontext index contracts/
fcontext index docs/architecture.docx

Supported formats

Category Formats
Documents PDF, DOCX, XLSX, PPTX, RTF, ODT, EPUB
Presentations Keynote, PowerPoint
Text Markdown, TXT, RST, AsciiDoc (direct copy)
Diagrams Excalidraw, Draw.io

Converted files are cached in .fcontext/_cache/ with source tracking. Incremental — only re-converts when source files change.

Dynamic Context Building

AI agents write conclusions and discoveries to _topics/ during sessions:

.fcontext/_topics/
  debugging-auth-flow.md       # Yesterday's debugging session
  architecture-decisions.md    # Key design choices
  api-integration-notes.md     # Third-party API findings

The next session automatically reads these files, continuing where the last session left off.

How topics work

Topics are plain Markdown files. The AI creates them when it reaches important conclusions. You can also write them manually. Use fcontext topic list to see what's accumulated.

Experience Packs

Package and share domain knowledge across projects and teams:

# Export your project knowledge
fcontext export team-knowledge.zip
fcontext export git@github.com:team/domain-knowledge.git

# Import into another project
fcontext experience import team-knowledge.zip
fcontext experience import git@github.com:team/domain-knowledge.git

# Keep packs up-to-date
fcontext experience update

An experience pack contains _README.md, _cache/, _topics/, and _requirements/ — everything an AI needs to understand a domain.

Real-world experience packs

Pack Description Source
fcontext-pmp-course Learn project management with AI fcontext experience import
fcontext-embed-development-course Learn embedded development with AI fcontext experience import
fcontext-domain-driven-design Learn DDD with AI fcontext experience import

Requirements Management

Track requirements, stories, tasks, and bugs with full evolution history:

fcontext req add "User authentication via OAuth" -t story
fcontext req add "Support Google provider" -t task --parent STORY-001
fcontext req set TASK-001 status in-progress
fcontext req board    # Kanban view
fcontext req tree     # Hierarchy view
graph TD R["ROADMAP-001<br/>Product Vision"] --> E["EPIC-001<br/>Auth System"] E --> S1["STORY-001<br/>OAuth Login"] E --> S2["STORY-002<br/>Session Mgmt"] S1 --> T1["TASK-001<br/>Google Provider"] S1 --> T2["TASK-002<br/>GitHub Provider"]

Requirements support evolution tracking — when requirements change, link the new version to the old one:

fcontext req add "OAuth with PKCE flow" -t story
fcontext req link STORY-003 supersedes STORY-001
fcontext req trace STORY-003   # see full history

Workspace Map

Auto-generated project structure overview in .fcontext/_workspace.map:

fcontext init    # generates workspace map automatically

The map gives AI agents an instant understanding of project layout without scanning every file.

Living Project Summary

.fcontext/_README.md is the first file every AI session reads. The AI maintains it automatically — updating it when it learns something significant about the project.

This creates a living document that captures:

  • Project purpose and domain
  • Architecture overview
  • Key conventions and patterns
  • Known pitfalls and decisions

Offline & Secure

  • All data local — stored in .fcontext/ directory
  • No cloud dependency — works without internet
  • No API keys required — fcontext itself needs no external services
  • No telemetry — zero data collection
  • Git-friendly — .fcontext/ is designed to be version-controlled