Context management
for AI agents
ContextAgora gives your agents access to the tools, services, and docs they need — without leaking secrets or losing control over what they can see.
Deploy guide →- Modular context
- Modules live in a GitHub repo and are symlinked into the agent workspace on demand. The agent only sees what you load.
- Runtime secrets
- Secrets are declared by name in a schema, never stored on disk. Varlock resolves values from Infisical at runtime and discards them when the command exits.
- Single container
- One Docker image, one
.envfile. Works with Anthropic, OpenAI, or Ollama. - Full audit trail
- Every change the agent makes is a git diff you can review, push, or discard.
┌─────────────────────────────────────────────────────┐
│ Web UI (:8080) │
│ │
│ Chat Modules Benchmarks │
│ talk to the create, edit, test if your │
│ AI agent load/unload context makes │
│ context the agent better │
└────────────────────┬────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────┐
│ Agent Workspace │
│ │
│ context/ │
│ ├── linear/ ← symlink ── modules-repo/linear │
│ ├── supabase/ ← symlink ── modules-repo/supabase │
│ └── CLAUDE.md │
│ │
│ The agent only sees what you've loaded. │
└──────────┬──────────────────────┬───────────────────┘
│ │
▼ ▼
┌──────────────────┐ ┌──────────────────────┐
│ GitHub Repo │ │ Infisical │
│ │ │ │
│ your modules │ │ secrets vault │
│ version- │ │ resolved at │
│ controlled │ │ runtime by │
│ │ │ varlock, never │
│ │ │ written to disk │
└──────────────────┘ └──────────────────────┘