Back to directory

Knowledge Agent Template

Knowledge Agent Template is an open-source, forkable starting point for an AI agent that answers questions from a knowledge base using grep, find and cat inside a sandbox, with no embeddings or vector database, deployable as a chat, GitHub or Discord bot.

Visit website (opens in a new tab)chatsdk-knowledge-agent.vercel.app
Knowledge Agent Template

Knowledge Agent Template is an open-source template for building a file-system and knowledge-based AI agent that stays current with its sources. Instead of an embeddings and vector-database pipeline, agents search across sources such as GitHub repositories, YouTube transcripts or custom APIs using grep, find and cat inside isolated sandboxes, producing deterministic and explainable results with no separate indexing infrastructure.

The same agent can be deployed as a web chat app, a GitHub Issues bot or a Discord bot through pluggable adapters, and it ships a built-in admin panel with usage statistics, error logs and source configuration, plus an AI-powered admin agent that answers operational questions such as recent errors or slowest endpoints. A complexity router sends simple questions to cheap models and hard ones to more capable models, and a pooled sandbox reuses already-running sandboxes to cut startup time.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.