Add a sentry mcp subcommand that launches the sentry-mcp server in stdio mode, automatically binding the access token from the CLI's existing auth stack — no manual token configuration required.
Current behavior
Users must run sentry-mcp as a separate tool and manually supply an access token. There's no path to launch the MCP server with credentials already managed by the CLI.
Gap
- MCP clients (Claude, Cursor, etc.) need a single shell command to wire up the Sentry MCP server. Wrapping it as
sentry mcp gives users a zero-config entry point that reuses their existing sentry login credentials.
- Without this, discoverability and setup friction are high — users have to know about
sentry-mcp as a separate package and obtain a token independently.
Implementation notes
- Can be done pre-repo merge by having
sentry-cli take a dependency on sentry-mcp (or exec it) and invoking it in --transport stdio mode.
- Auth binding should follow the same resolution order the CLI already uses: stored credentials →
SENTRY_AUTH_TOKEN env var.
- MCP client config would simplify to:
{ "command": "sentry", "args": ["mcp"] }.
--
View Junior Session in Sentry
Add a
sentry mcpsubcommand that launches the sentry-mcp server in stdio mode, automatically binding the access token from the CLI's existing auth stack — no manual token configuration required.Current behavior
Users must run
sentry-mcpas a separate tool and manually supply an access token. There's no path to launch the MCP server with credentials already managed by the CLI.Gap
sentry mcpgives users a zero-config entry point that reuses their existingsentry logincredentials.sentry-mcpas a separate package and obtain a token independently.Implementation notes
sentry-clitake a dependency onsentry-mcp(or exec it) and invoking it in--transport stdiomode.SENTRY_AUTH_TOKENenv var.{ "command": "sentry", "args": ["mcp"] }.--
View Junior Session in Sentry