DEDUPED Hot AI repos, turned into use cases Get the Monday 5

How to index private code on your own CPU with no token spend with Heimdall

Build a searchable memory of your repos that never leaves your machine.

Who it is forA developer with private or client code who wants searchable agent memory without sending files to a cloud model.

The problemMost memory tools send code to an extraction LLM or charge tokens for every indexed file.

The outcomeA local index built with tree-sitter parsing and local embeddings, with search running against your local daemon.

What you need first

  • Node.js with npm, to run `npm i -g @arihantdeva/heimdall`
  • macOS or a Debian, Ubuntu, Fedora or Amazon Linux machine, with the build prerequisites listed in the README for the graft backend
  • git, because llama.cpp is cloned at first build
  • A supported agent harness: pi, Claude Code, Codex, Cursor, Windsurf, or opencode

Steps

  1. npm i -g @arihantdeva/heimdall
  2. heimdall setup --detect-only
  3. heimdall setup --model bge-small-en-v1.5
  4. heimdall doctor
  5. heimdall search "<topic>" (inferred, check the README)

Prompt for your agent

Using the ArihantDeva/heimdall CLI, run `heimdall setup --detect-only` and report the hardware profile, then run `heimdall doctor` and confirm it prints HEALTHY. Expected output: the hardware profile and the doctor status. Verify that no network call is needed for indexing by checking the README's stated local-only design.

Caveat. The README says the Linux daemon packaging is still unchecked on the roadmap, so the background daemon setup is documented for macOS launchd and may need manual work on Linux.

Use this in your AI agent

Ask your agent

Using the Deduped MCP, get the setup steps for ArihantDeva/heimdall and "Index private code on your own CPU with no token spend"

Add Deduped to Claude Code once

claude mcp add --transport http deduped https://dedupedai.com/mcp

Cursor, ChatGPT, others

Share: X LinkedIn Hacker News Email Copy link

Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer