Dosu pricing, free plan and limits

Last updated: 2026-09-27

Dosu is a knowledge-infrastructure platform for AI coding agents: it observes how a team's engineers and its connected agents work, then shares that context over MCP so every agent benefits from what one already learned. It also keeps documentation synced automatically as code changes, with over 5,955 repositories connected.

About Dosu

Dosu is an AI-powered engineering knowledge platform built by founder Devin Stein, who launched the company in San Francisco in 2023 as a GitHub bot for automated issue triage. It raised $8.52M in a Series A round in April 2024, led by Sequoia Capital with participation from Innovation Endeavors, Materialized View Capital, and Rsquared.

The product has since repositioned around what Dosu calls its "knowledge infrastructure for agents." It watches how engineers and AI coding agents work across GitHub, Slack, Notion and Confluence, then shares that context with every connected agent over the Model Context Protocol, including Claude Code and GitHub Copilot. It auto-generates AGENTS.md files and keeps documentation in step with the codebase, rather than sitting on a manual doc backlog.

For open-source maintainers, the free plan still applies to public repositories, and the auto-labeling and stale-issue bots that used to run as hosted GitHub features are now open-sourced as standalone GitHub Actions, auto-label and better-stale-bot, rather than discontinued. Open-source projects backed by CNCF or the Apache Software Foundation get a partner plan with direct Slack and Discord access to the Dosu team. More than 5,955 repositories use Dosu today.

Dosu is not a fit for teams that want free, unlimited private-repo automation: the old Community-tier policy of unlimited monthly requests on any public repo is gone, replaced by a metered, public-repo-only free plan, with private-repo access requiring a paid upgrade. Teams that only want the classic GitHub issue-triage bot, without the MCP knowledge layer feeding agents like Cursor, may prefer running the newly open-sourced auto-label and better-stale-bot Actions directly. Greptile takes a different approach entirely, indexing a whole codebase for pull-request review. Qodo rounds out the field with test generation bolted onto code review, and Pancake folds light triage into a wider cofounder product.

Pricing

The Free tier costs nothing and includes 200 credits a month, but is now restricted to public repositories only; the old unlimited-request policy for open source is gone. Pro costs $16/month ($192 billed annually) for 4,000 credits and adds private-repo access plus Slack, Confluence, Coda, Notion and Linear integrations. Enterprise pricing is quoted through a sales call and adds SSO, custom RBAC and audit logs.

A separate Maintainers Pro tier, also custom-priced, is open to OSS projects backed by a major organization or VC funding. Browse other AI documentation generators on HokAI for lower-cost alternatives if the credit model doesn't fit.

Key Features

• Automatic Knowledge Capture: Watches commits, pull requests, Slack threads and Notion or Confluence pages, then reconciles the facts into a shared knowledge base without a manual documentation step.
• MCP-Native Agent Memory: Connects to Claude Code, Cursor, GitHub Copilot and Windsurf over the Model Context Protocol, so what one agent learns becomes available to every other agent on the team.
• Automated Documentation Pull Requests: Opens a documentation update as its own pull request whenever a code change leaves AGENTS.md, CLAUDE.md or other agent-facing docs out of sync.
• Credit-Based Usage Model: The free plan meters usage in credits and applies only to public repositories; upgrading to the paid Pro plan raises the monthly allowance and adds private-repo access.
• SaaS Integrations: Pro and Enterprise plans connect Slack, Confluence, Coda, Notion and Linear so knowledge captured across tools feeds the same context layer.
• Open-Sourced GitHub Triage Tools: auto-label and better-stale-bot, the original hosted issue-triage and stale-issue features, now run as free, self-hosted GitHub Actions instead of a paid service.

Pros

• The free plan gives any public open-source repository a monthly credit allowance for MCP-based agent documentation, no card required.
• Knowledge captured by one coding agent, such as Cursor, Claude Code or Copilot, is shared with every other agent on the team, cutting the repeated context-gathering that drives up token spend.
• A CNCF or Apache Software Foundation affiliation is enough to reach the extended partner plan, which throws in closer Slack and Discord contact with the team behind the product.
• The formerly hosted auto-labeling and stale-issue bots were open-sourced rather than shut down, so free-plan users who relied on them are not left with nothing.

Cons

• The unlimited-request open-source model is gone: the free plan now meters usage on a monthly credit allowance and only covers public repositories, a real step down from the previous request-based policy.
• Private-repository access requires upgrading to the paid Pro plan at minimum, and Teams- or Enterprise-level SaaS integrations and SSO are quoted through a sales conversation rather than listed.
• The product's identity shifted from a GitHub issue-triage bot to a cross-tool agent knowledge layer, so teams that specifically want hosted auto-labeling now have to run the open-sourced Action themselves.

Visit Dosu Official Website