Browse / LLM wiki / agent-wiki

agent-wiki

agent-observatory

A personal knowledge space for you and your agents.

Open agent-observatory/agent-wiki → Does not match the four rules.

Works with any agent

ships AGENTS.md, the cross-tool instruction file.

In its own words

README.mdagent-observatory/agent-wiki

A personal knowledge space for you and your agents. Review decisions, preserve evidence, and manage knowledge through your agent and CLI. The web is a read-only viewer.

  • Workspace isolation, GitHub owner login, scoped agent API keys
  • Keyword and glossary-alias search; no embedding dependency
  • Background text refinement with source spans and revision lineage
  • Next.js, Fastify, PostgreSQL, K3s, Terraform

First release deployed: agent-wiki.duckdns.org. Owner-only GitHub login. Source collection, knowledge APIs, and URL-based pages are live. See implementation and deployment status.

A read-only Collector and remote AI Worker run outside the active conversation. BYOK settings control the provider, model, credentials and optional limits through the CLI. The agent-wiki-client package in packages/agent-wiki-client includes agent-wiki-cli, agent-wiki-collector and the Agent Wiki Skill. The Skill guides retrieval and user-authorized review, the CLI executes queries and management commands, and the Collector runs independently. See the installation guide below. Development mode allows a clean data reset; backward compatibility is not required.

Development

Node.js 22.21+, Docker and PostgreSQL 17 are required.

Quoting 80% of the README. Read the rest on GitHub ↗

Read from the repository's git tree on 2026-08-17.