Who it is forAn engineer who wants an agent to answer questions about a large codebase or docs set.
The problemFlat vector search over everything returns noisy matches from unrelated parts of the project.
The outcomeThe repo indexed under viking://resources with abstracts and overviews, searchable by subtree.
What you need first
- uv and Python for a self-hosted server, or the hosted OpenViking Service
- A model provider for the server (Volcengine, OpenAI, Codex OAuth, Kimi, GLM or local Ollama)
- A user key for the memory plugin (a root key cannot read or write memories)
Steps
- With the server running, run
ov add-resource https://github.com/volcengine/OpenViking(or your repo) - Check progress with ov task status and your task ID
- Browse with
ov ls viking://resources/andov tree viking://resources/volcengine -L 2 - Search with
ov find "what is openviking"or grep inside one folder with --uri
Prompt for your agent
Use volcengine/OpenViking's ov CLI to import <repo URL>, wait for the task to finish, then answer <question> using ov find scoped to the right directory and show the URIs you used.
Caveat.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for volcengine/OpenViking and "Import a repository and search it one directory at a time"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer