Standalone apps that do the agent work themselves. Knowledge graphs, extraction platforms, workflow orchestrators, coding-agent runtimes — you install and run them, they perform the agent tasks. Not libraries you code against.
8 apps
Self-hosted AI knowledge assistant (formerly Danswer). Connects to your team's Slack, Google Drive, Confluence, GitHub and turns them into a Q&A agent with source citations. The default open-source Glean alternative.
Client-side knowledge graph creator for codebases with a built-in Graph RAG agent. Runs entirely in the browser — no server. Point it at a repo and get an interactive graph plus a chat agent that answers with graph-aware context.
LLM-driven document extraction platform. Turn any PDF, image, or scanned doc into structured JSON with prompt-based extraction. Ships an agent workflow builder and cloud-hostable API. Common alternative to LlamaParse and Unstructured.
Self-hosted workflow orchestration with first-class support for AI agent steps. Airflow alternative with a lighter footprint — single Go binary, YAML DAGs, built-in web UI. Runs LLM tasks alongside shell, HTTP, and Docker steps.
Code-intelligence engine for AI agents. Builds a semantic index of your codebase that agents can query for symbol lookup, call-graph traversal, and refactor planning. Runs as a local service; language-model-agnostic.
Turn your AI IDE session history (Cursor, Windsurf, Claude Code) into reusable skills and specs. Runs locally, watches your IDE, and generates SKILL.md-style artifacts from what you actually did.
Local-first control plane for ComfyUI with agent workflows. Exposes ComfyUI graphs to LLM agents via MCP so an agent can compose image workflows, run them, and read back the results.
Memory operations app for coding agents. Persists and retrieves cross-session context so your agent remembers what it did in previous sessions on the same repo. Sits alongside your harness (Claude Code, Cursor, etc.).