Skip to content

docs: describe MCP OAuth tool provider flows#630

Draft
cursor[bot] wants to merge 1 commit intomasterfrom
cursor/automated-documentation-updates-4827
Draft

docs: describe MCP OAuth tool provider flows#630
cursor[bot] wants to merge 1 commit intomasterfrom
cursor/automated-documentation-updates-4827

Conversation

@cursor
Copy link
Copy Markdown
Contributor

@cursor cursor bot commented Apr 16, 2026

Summary

Documented the MCP OAuth behavior introduced by the merged master change set from PR #621.

Triggering code changes documented here:

  • Added per-user MCP OAuth state in the MCP tools response (needsOAuth / authURL) so the webapp can show connection state.
  • Added plugin-managed MCP OAuth start and disconnect flows for individual users.
  • Updated the Agents RHS tool-provider UI to show Connect and Disconnect for OAuth-backed MCP servers.
  • Removed server-posted ephemeral OAuth prompts from conversations.
  • Added same-origin validation for resource_metadata during MCP OAuth start.

Documentation files updated:

  • docs/admin_guide.md — Added guidance for OAuth-backed MCP servers, including the per-user connection model, System Console behavior, current web/desktop-only connection flow, and the resource_metadata origin requirement.
  • docs/user_guide.md — Added end-user instructions for the Agents pane Tools menu, MCP provider Connect / Disconnect behavior, saved provider preferences, and the current mobile limitation.

PR context incorporated from #621:

  • The docs note that ephemeral OAuth prompts were intentionally removed and users should use the Agents webapp Tools menu instead.
  • The docs note that MCP OAuth connections are currently initiated from Mattermost web or desktop, not mobile.

Screenshots:

  • N/A. This is a docs-only change and no screenshots were required or captured.

Borderline items for review:

  • None.

Ticket Link

N/A

Screenshots

N/A

Release Note

NONE
Open in Web View Automation 

Co-authored-by: Nick Misasi <nick13misasi@gmail.com>
@github-actions
Copy link
Copy Markdown

🤖 LLM Evaluation Results

OpenAI

⚠️ Overall: 18/19 tests passed (94.7%)

Provider Total Passed Failed Pass Rate
⚠️ OPENAI 19 18 1 94.7%

❌ Failed Evaluations

Show 1 failures

OPENAI

1. TestReactEval/[openai]_react_cat_message

  • Score: 0.00
  • Rubric: The word/emoji is a cat emoji or a heart/love emoji
  • Reason: The output is the text "heart_eyes_cat", not an actual cat emoji (e.g., 😺/🐱) or a heart/love emoji (e.g., ❤️/😍). Therefore it does not meet the rubric requirement.

Anthropic

⚠️ Overall: 18/19 tests passed (94.7%)

Provider Total Passed Failed Pass Rate
⚠️ ANTHROPIC 19 18 1 94.7%

❌ Failed Evaluations

Show 1 failures

ANTHROPIC

1. TestReactEval/[anthropic]_react_cat_message

  • Score: 0.00
  • Rubric: The word/emoji is a cat emoji or a heart/love emoji
  • Reason: The output is the text "heart_eyes_cat", which is not itself a cat emoji or a heart/love emoji character.

This comment was automatically generated by the eval CI pipeline.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant