Optiq Code — Deterministic Code Context Engine for AI Agents

Optiq Code is a deterministic context retrieval engine for software repositories. It parses source code into abstract syntax trees (ASTs), builds dependency graphs, and creates hybrid search indexes combining semantic vector search with keyword matching. When an AI coding agent needs to find relevant code, Optiq returns ranked results with function signatures, call graphs, and confidence scores — in approximately 1-2 seconds.

Key Features

• Hybrid Search: Semantic vector search + keyword matching (Tantivy) in a single query
• AST Parsing: Full abstract syntax tree analysis for 12 languages — Python, TypeScript, JavaScript, TSX, Rust, Go, Java, C, C++, Ruby, PHP, C#
• Symbol Resolution: Tracks imports, exports, function calls, and references across files
• Confidence Scoring: Every result includes a relevance score (0-1) so agents can filter by quality
• MCP Server: Built-in Model Context Protocol server for direct AI agent integration
• CLI Tool: npm install -g @optiqcode/cli — index, search, and serve from the terminal
• Security: AES-256-GCM encryption at rest, per-user database isolation, 99.9% SLA

When to Use Optiq

Use Optiq when your AI coding agent needs to find the right function in a large codebase without scanning every file, understand how code connects across modules, get precise context windows instead of dumping entire files into the prompt, or reduce hallucination by providing verified, ranked code context.

How It Works

1. Index: Point Optiq at your repository. It parses every file into an AST, extracts symbols, and builds a dependency graph.
2. Search: Query with natural language or structured filters. Optiq combines semantic and keyword search to find the most relevant code.
3. Retrieve: Get ranked results with function signatures, confidence scores, and cross-file context — ready for your AI agent's prompt.

API

Base URL: https://api.optiqcode.com

Full API specification: OpenAPI 3.1 spec

Authentication

All API requests require a Bearer token. Get your API key from the dashboard (self-serve, no credit card required). Keys support scoped permissions: search:read, repos:read, repos:write, repos:delete, jobs:read.

OAuth Discovery

Authorization server metadata: /.well-known/oauth-authorization-server. Supports PKCE with S256 code challenge method.

Rate Limits

Every API response includes rate limit headers: X-RateLimit-Limit, X-RateLimit-Remaining, X-RateLimit-Reset, and Retry-After (on 429 responses).

Streaming

Add Accept: text/event-stream for Server-Sent Events (SSE) during indexing and search. Events include progress updates and streamed results with confidence scores.

Error Handling

All errors return structured JSON: {"error": {"code": "...", "message": "...", "hint": "..."}}. HTTP status codes: 400 (bad request), 401 (unauthorized), 403 (forbidden scope), 404 (not found), 429 (rate limited), 500 (server error).

MCP Server Integration

Optiq exposes an MCP server for direct AI agent integration.

Discovery

• MCP manifest: /.well-known/mcp
• Server card: /.well-known/mcp/server-card.json
• Remote endpoint: https://api.optiqcode.com/mcp (Streamable HTTP)
• Local: optiq mcp (stdio transport)

MCP Tools

Pricing

Machine-readable pricing: /pricing.md

Installation

npm install -g @optiqcode/cli
optiq login
optiq index .
optiq search "your query"

Supported Languages

Python, TypeScript, JavaScript, TSX, Rust, Go, Java, C, C++, Ruby, PHP, C#

Integrations

Works with Cursor, Windsurf, VS Code, JetBrains, Neovim, Claude Desktop, Claude Code, and any MCP-compatible client.

Compare

See how Optiq compares to grep, GitHub Code Search, Sourcegraph, and embedding-only RAG.

Machine-Readable Discovery

• OpenAPI 3.1 spec
• MCP server manifest
• Agent card
• OpenAI plugin manifest
• OAuth metadata
• API catalog (RFC 9727)
• Protected resource (RFC 9728)
• Pricing (markdown)
• llms.txt
• llms-full.txt
• Agent mode (JSON)