Optave Releases Code Intelligence CLI with AI Agent Focus

What changed

The ops-codegraph-tool, developed by Optave, is a new command-line interface (CLI) designed to provide deep code intelligence. It constructs a function-level dependency graph for entire codebases, supporting 11 languages. The tool utilizes tree-sitter for code parsing and stores the graph in SQLite, making it accessible via an MCP (Multi-Cloud Platform) server for AI agents, a CLI for developers, and a programmatic API. This approach aims to give AI agents a comprehensive understanding of code structure, preventing common issues like modifying functions without awareness of their callers or leaving behind dead code. The tool emphasizes local execution with zero API keys or telemetry, ensuring code privacy. Key features include generating function-level dependency graphs, providing complexity metrics, enforcing architecture boundaries, and integrating with CI quality gates. It also offers Git diff impact analysis with co-change analysis and hybrid semantic search capabilities. The latest release mentioned is dev-v3.11.3-dev.4. The tool is licensed under Apache-2.0.

Why it matters for builders

For AI builders, ops-codegraph-tool offers a solution to a significant problem: AI agents often lack the necessary context to navigate and modify codebases effectively. By providing a detailed, function-level dependency graph, the tool equips agents with structural awareness, reducing the likelihood of errors stemming from misunderstandings of code relationships. This can lead to more reliable code generation and modification by AI agents. Developers can also benefit from enhanced code exploration, auditing, and the enforcement of architectural standards directly within their workflow.

Practical impact

Developers can integrate ops-codegraph-tool into their workflow by installing it globally via npm (npm install -g @optave/codegraph). After navigating to their project directory, they can build the code graph with a simple codegraph build command. This process requires no configuration files, Docker, JVM, API keys, or accounts. Once built, the MCP server can be pointed at by an AI agent, granting it immediate structural awareness of the codebase. The CLI offers commands like context handleAuth -T to explore dependencies and callers. For CI pipelines, check --staged can be used to enforce quality thresholds and architectural rules, failing builds if complexity or blast radius thresholds are exceeded. The manifesto and boundaries commands are specifically designed to enforce architectural rules on every commit.

Caveats and source limits

The provided source material is primarily the GitHub repository README. While it details the capabilities and intended use of ops-codegraph-tool, it does not include independent benchmark results comparing its performance or accuracy against other code intelligence tools. Specific pricing information is not available as the tool is open-source and free. The source also mentions a "fresh release" but does not provide a specific release date beyond the latest version tag. The comparison table indicates that while the tool supports 11 languages, other tools may support more, and the depth of certain features like interprocedural dataflow or Git-aware analysis might vary. The source explicitly states that security scanning (SAST/vulnerability detection) is intentionally out of scope for this tool.