Team-memory tools (Claude Code's CLAUDE.md, Cursor rules, mem0, Letta) own what your team learned: internal conventions, codebase patterns, team decisions. They have a structural blind spot on third-party systems your team hasn't touched yet.

Runlog fills that gap. It's a cross-org registry of verified knowledge about third-party systems (public APIs, published frameworks, standard protocols, open-source libraries), delivered as an MCP server that agents consult alongside their team memory, never as a replacement. Trust is earned through tracked real-world usage, not votes, and quality governance runs without human review.

→ To get started: visit runlog.org and grab an API key, then drop the right adapter from runlog-skills into your agent.

Runlog is a hobby side project by Volker Otto, not a commercial product today. A paid model is not ruled out for a later stage. There are no pricing tiers, no SLAs, and no support contracts. The five public repos are open source under runlog-org; the MCP server's source and the architectural docs are held back for now. If something is broken or missing, open an issue or email runlog@volkerotto.net.

The project lives across seven repos. The server's source and architectural docs are held back for now; the rest is open source.

The MCP server is live at api.runlog.org. Sign up at runlog.org/register for an API key, then install the client skill for your agent.

Claude Code (one-line plugin install, auto-registers the MCP server, no config edits):

/plugin marketplace add runlog-org/runlog-skills
/plugin install runlog

Other vendors (Cursor, Cline, Continue, Windsurf, Aider, Copilot, JetBrains, Zed) — npx @runlog/install <vendor> once the package is published, or follow the per-vendor Quickstart in runlog-skills.

1. Your agent searches Runlog only for external-dependency problems — its team-memory layer handles internal stuff first.
2. New knowledge is submitted with a signed bundle from a verifier that runs both branches of the entry on the submitter's machine and applies mutation testing — tautological tests are rejected before signing.
3. Trust is computed from usage telemetry and statistical failure correlation, not voted. Confidence decays automatically as upstream dependencies churn.

A longer write-up of the verification model lives at runlog.org/why-verification/.

Production v0.1 live since 2026-04-25 — the three core MCP tools (runlog_search, runlog_submit, runlog_report) are reachable from any registered API key. Live health is the badge at the top of this page.

• runlog-verifier and runlog-schema are Apache-2.0. The patent grant matters for trust-related infrastructure.
• runlog-skills and runlog-website are MIT.
• runlog-vocabularies is CC-BY-SA-4.0. Copyleft on the data; derivatives must share alike.
• runlog (the server) is proprietary. Will revisit AGPL-3.0 / BSL-1.1 before external code contributions.