Warp: GitHub PR Comments Skill and Cross-Tool MCP Server Discovery
WarpWarp's March 25, 2026 release introduced the /pr-comments skill, enabling AI agents to fetch Pull Request comments directly from GitHub within a terminal session, eliminating context-switching to a browser for code review feedback. The release also added cross-tool MCP server discovery: MCP servers installed by third-party AI tools like Claude and Codex are now automatically visible and spawnable from Warp's AI settings, reducing configuration duplication across the AI development toolchain.
New GitHub Integration: /pr-comments Skill
Warp introduced a new built-in skill — /pr-comments — that allows AI agents to retrieve Pull Request comments from GitHub directly within a terminal session. For developers who iterate on code review feedback, this means the agent can now read PR comments as context without requiring a browser switch or manual copy-paste of reviewer notes.
The skill integrates natively into Warp's agent input flow, invokable with a slash command during an active conversation. This extends Warp's growing library of GitHub-aware capabilities, which already included branch and diff awareness. With /pr-comments, a developer can ask the agent to address specific reviewer feedback by referencing the PR, and the agent retrieves that data on demand.
A related improvement allows code review comments to be sent directly to a running CLI agent. A developer can highlight a reviewer comment and forward it as a prompt to an active agent already working in the same context — no interruption to the current workflow required. A GitHub Pull Request chip also now appears in AI input mode when a PR exists for the current branch, giving agents passive awareness of ongoing code review from the moment a session starts.
Cross-Tool MCP Server Discovery
Warp now automatically detects and surfaces MCP (Model Context Protocol) servers that were installed and configured by other AI tools on the same machine — specifically Claude and Codex. These servers appear in Warp's AI settings page under the MCP servers section, where they can be inspected and spawned directly.
Previously, developers using multiple AI tools would need to manually duplicate MCP server configurations in each tool's settings. With this change, Warp acts as an aggregation layer for the MCP ecosystem: it recognizes servers set up by peer tools and makes them immediately available without redundant configuration. This is particularly valuable for teams standardizing on a shared set of MCP servers across different AI agents in their workflow.
Minor Additions
A new /changelog command was added that reopens the latest changelog entry directly in Warp, keeping release notes accessible during a session without opening external docs.