CTX

Overview

CTX is a Rust-based local runtime layer that serves as the context engine for OpenCode and AI coding agents. It indexes repositories, stores reusable guidance as graph memory, exposes local MCP tools, and maintains telemetry about token savings. CTX addresses token waste by implementing strategies like graph memory for reusable rules, read-cache compression for file rereads, command pruning for logs, and compact context packs. Unlike agent launchers, CTX integrates directly into the OpenCode workflow as a context layer.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• query

  Query repository knowledge with hybrid graph, snippets, FTS, symbols, and semantic ranking

  Try:Find all functions related to authenticationTry:Search for database connection patterns
• read_file

  Read files with compressed modes: full, outline, or digest

  Try:Read main.py in digest modeTry:Show outline of config.json
• pack

  Build compact task packs with graph, memory, failure, diff, and attachment signals

  Try:Pack context for user authentication featureTry:Prepare task pack for database migration
• plan

  Create implementation plans by combining retrieval, graph, memory, and pack signals

  Try:Plan implementation of user dashboardTry:Create development roadmap for API integration
• run_command

  Execute shell commands with root cause extraction and log storage

  Try:Run tests and extract key findingsTry:Execute deployment script and analyze output
• learn

  Store reusable project lessons in graph memory

  Try:Learn best practices for error handlingTry:Store authentication flow patterns
• toolbook_import

  Import toolbooks into CTX for large CLI manuals

  Try:Import Docker CLI guideTry:Add AWS command reference
• toolbook_search

  Search imported toolbooks for specific information

  Try:Find Docker build commandsTry:Search for AWS S3 operations
• toolbook_pack

  Pack relevant toolbook sections for specific tasks

  Try:Pack Kubernetes deployment commandsTry:Prepare bash script reference
• prune_logs

  Prune logs and shell output to extract root cause signals

  Try:Prune build logs for errorsTry:Extract key information from deployment output
• get_stats

  Retrieve token savings and performance metrics

  Try:Show token savings summaryTry:Get recent performance metrics
• dashboard

  Display live CTX dashboard with local runtime metrics

  Try:Show me the CTX dashboardTry:Display live metrics

Note: Tool names inferred from slash commands documented in the OpenCode usage section. The documentation mentions MCP functionality but doesn't explicitly list the MCP tool names or signatures.

Comparable tools

Installation

# Via Cargo
cargo install ctx-cli

# One-line installer
curl -fsSL https://raw.githubusercontent.com/Alegau03/CTX/main/scripts/install.sh | sh

# Via npm
npm i -g @alegau/ctx-bin

# Via Homebrew
brew tap Alegau03/ctx && brew install ctx

cd /path/to/your/project
ctx init
ctx index
ctx opencode install
opencode

{
  "mcpServers": {
    "ctx": {
      "command": "ctx",
      "args": ["mcp"]
    }
  }
}

FAQ

What makes CTX different from other AI coding tools?

CTX is not another agent launcher but a context runtime layer that optimizes how information is passed to AI models through compression, caching, and graph memory, reducing token waste while preserving the original AI model and workflow.

How does CTX reduce token usage in AI coding workflows?

CTX implements several strategies including graph memory for reusable rules, read-cache compression for file rereads, command pruning for logs, delta-aware indexing, and compact context packs that contain only task-relevant information.

Compare CTX with