Agents

Manage your AI agents — each gets its own provider, model, and MCP tools

Adding or updating an agent requires an app restart to take effect. Use the Restart App button in the sidebar once you're done.

Agent Runs

Long-running agent tasks — start one, then track it here without waiting on an open connection

Feedback Loop

Turn 👍/👎-rated agent runs into a tuned system prompt — review and test-drive before anything is applied

👍 0 👎 0 0 unrated
Rated runs — pick which ones to learn from
Revision History

A2A Agents

Compose existing agents into sequential or concurrent multi-agent workflows

Workflows

Build multi-agent workflows (Sequential, Concurrent, Handoff, GroupChat, Magentic) — each is exposed as an agent

Skills

Build reusable skill modules — instructions, reference resources, and callable scripts

MCP Servers

Connect to Model Context Protocol tool servers

Adding, updating, or removing an MCP server requires an app restart to take effect. Use the Restart App button in the sidebar once you're done.

RAG Connections

External vector store connections used by knowledge bases for retrieval-augmented generation

Knowledge Bases

Ingest documents for retrieval-augmented generation — assign to agents in the Knowledge tab

Agent Files

Files Harness agents have written via File Memory or File Access — stored in Postgres, browse and edit them here

Graph Connections

Connections to a graph store — Apache AGE runs inside this app's own Postgres (no extra server); Neo4j points at a separate server

Relational Connections

External Postgres sources used by the "Import from Database" wizard on a Knowledge Graph — schema, tables, and rows are read-only; the app never writes to these

Strapi Connections

CMS sources RSS/Atom feeds are built from. Add the API token afterwards, from Secrets.

RSS/Atom Feeds

Feeds built from Strapi collections. Consumers poll the public URL for approved, unfetched items — a successful poll marks them delivered.

Knowledge Graphs

Graphs agents can retrieve from — assign them to an agent in its Graph tab

Providers

Add any number of LLM providers — OpenAI, Anthropic, Gemini, Ollama, or any compatible endpoint

Adding, updating, or removing a provider requires an app restart to take effect. Use the Restart App button in the sidebar once you're done.

Secrets

Manage API key secrets stored in the database — providers reference these by name

Values are stored in plain text. For production, use environment variables prefixed with Secrets__ — they take priority over DB values. Adding, updating, or removing a secret requires an app restart to take effect. Use the Restart App button in the sidebar once you're done.

Users

Dashboard accounts — Admins can also manage users; Members have full access otherwise

API Reference

All endpoints exposed by this deployment — click any row to copy the full URL

Interactive Docs
Base URL
🤖
Select or start a conversation
Choose an agent, then create a new conversation to begin chatting.