Hippocortex/Docs
GitHubDashboard

Onboarding: Claude Code

Add persistent memory to Claude Code. Your coding agent will recall past debugging sessions, deployment procedures, and project-specific knowledge across sessions.

Best path: Gateway (change one URL) Time: 2 minutes Reliability: ~99%

Step 1: Get an API key

Sign up at dashboard.hippocortex.dev. Copy your API key (hx_live_...).

Step 2: Configure the Gateway

Set environment variables so Claude Code routes through the Hippocortex Gateway:

export ANTHROPIC_BASE_URL="https://api.hippocortex.dev/v1"
export ANTHROPIC_API_KEY="hx_live_..."
export HIPPOCORTEX_LLM_API_KEY="sk-ant-..."  # Your real Anthropic key

Or add to your shell profile (~/.zshrc / ~/.bashrc) to persist across sessions.

Step 3: Use Claude Code normally

Every LLM call now passes through the Gateway. Memory is automatic:

  • Past conversations are captured and compiled into knowledge
  • Relevant context is injected into every new request
  • Debugging patterns, deployment procedures, and project knowledge persist across sessions

No tools to call, no prompting needed. The Gateway handles the full pipeline: capture, synthesize (semantic search, graph retrieval, collective brain), learn, and vault.

Step 4: Verify

After a few sessions, check dashboard.hippocortex.dev to see captured events and compiled artifacts.

The Gateway has ~99% reliability with graceful fallback — if the memory layer is temporarily unavailable, requests pass through directly to Anthropic so Claude Code never breaks.