Skip to content

Home

Why ContextAgent⚓︎

  • Context is the API — define prompts, memory, and tool state once and reuse across any agent.
  • Pipeline-first ergonomics — compose async steps with tracing, retries, and analytics built in.
  • Bring your own models — mix and match OpenAI, Claude, Gemini, DeepSeek, and self-hosted models per agent.
  • Production backed — audit logs, artifact storage, and MCP integration to ship safely.

What you can build⚓︎

Popular scenarios

  • Research teammates that synthesize web findings into structured briefs.
  • Workflow copilots that translate requirements into actions across your SaaS stack.
  • Data science orchestrators that run explorations, modeling, and reporting end-to-end.

Try it in minutes⚓︎

uv sync
uv run python -m examples.web_researcher

uv run contextagent --help

Architecture at a glance⚓︎

graph TD
  A[Pipeline Entry] -->|Query| B[Context Core]
  B --> C[Agent]
  B --> D[Tool Bus]
  C --> E[LLM Provider]
  D --> F[Integrations]
  B --> G[Artifacts]
  G --> H[Observability]