One context layer. MCP, REST, or both.

The same indexed codebase your team queries, wired into Cursor, Claude Code, Continue, Cline, or anything that speaks HTTP.

Get Started Free Read the Docs

terminal · zsh

# Add CodeAlive MCP to Claude Code
claude mcp add --transport http codealive \
  https://mcp.codealive.ai/api/ \
  --header "Authorization: Bearer $CODEALIVE_KEY"

# Or via REST
curl -X POST https://api.codealive.ai/v1/search \
  -H "Authorization: Bearer $CODEALIVE_KEY" \
  -d '{"query":"How does refund flow work?"}'

Drop-in MCP + REST

Connect your agent in one command

Connect any MCP-compatible AI agent to your indexed codebase in one command. No plugins, no glue code.

Works with Cursor, Claude Code, Continue, and Cline out of the box

No prompt templates, no glue code, no plugin to maintain

Reduces hallucinations by grounding answers in real code

Maintains full codebase awareness across every turn

Add CodeAlive to Claude Code over remote HTTP MCP:

claude mcp add --transport http codealive https://mcp.codealive.ai/api --header "Authorization: Bearer YOUR_API_KEY"

Plain HTTP. Bearer token. JSON.

Search, chat, stream tokens, batch jobs. Same layer your agents hit, exposed as REST.

RESTful endpoints

Predictable JSON over HTTPS. Bearer token auth. Works from any language or runtime.

Search & query

Hybrid semantic + lexical search, plus structured chat queries grounded in your code.

Streaming responses

Server-sent events for low-latency token streaming straight into your UI.

Batch operations

Submit many queries in one call, ideal for offline analysis and bulk pipelines.

Search across your repos with a single REST call:

curl https://api.codealive.ai/v1/search \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query": "rate limiting middleware", "data_source_ids": ["..."]}'