add managed auth tools (manage_auth_connections, manage_credentials, manage_credential_providers) by masnwilliams · Pull Request #104 · kernel/kernel-mcp-server

masnwilliams · 2026-05-24T19:56:49Z

Summary

Closes the largest agent-facing capability gap in the MCP server: setting up an authenticated browser session for a third-party site. Agents can now drive Kernel's managed auth flow end-to-end without a human in the loop (beyond the one-time hosted login).

What's added

manage_auth_connections (full surface)
- create — start managing auth for a profile + domain (optionally referencing a pre-stored credential by name, or an external provider like 1Password)
- list / get / delete
- login — kicks off a hosted login flow. Returns hosted_url (share with the user to sign in) and live_view_url (agent can watch). Triggers automatic re-auth if credentials are saved.
- submit — provide field values, an MFA option ID, or an SSO button selector when the flow is awaiting_input. Agent inspects discovered_fields / mfa_options from get to know what's needed.
manage_credentials (read-only)
- list / get (SDK never returns values) / totp_code (current 6-digit code)
- Intentionally no create/update/delete — credentials are stored by humans via dashboard/CLI so raw secrets never enter the agent's context. The agent references creds by name.
manage_credential_providers (read-only)
- list / get for external providers (e.g. 1Password). Same human-creates / agent-consumes pattern.

Agent flow

Human pre-stores netflix-mason credential in the Kernel dashboard.
Agent: manage_auth_connections create domain=netflix.com profile_name=mason credential_name=netflix-mason
Agent: manage_auth_connections login id=<conn_id> → shares hosted_url with user, or proceeds via re-auth.
Agent polls manage_auth_connections get until flow_status=SUCCESS (or flow_step=AWAITING_INPUT for MFA).
If MFA needed: manage_credentials totp_code <name> → manage_auth_connections submit fields={mfa_code: "123456"}.
Any future manage_browsers create profile_name=mason gets a logged-in session.

Test plan

Local server starts; tools register without errors
manage_auth_connections list returns existing connections for the authed user
manage_auth_connections create + login returns a working hosted URL
manage_credentials list returns names; totp_code returns a 6-digit code for a TOTP-enabled credential
manage_credential_providers list returns configured providers (or empty list)

Tool count

Bumps from 10 → 13.

Note

High Risk
Introduces auth flows, credential CRUD, TOTP retrieval, and provider tokens—security-sensitive surfaces where misuse or SDK behavior could expose secrets or weaken login boundaries.

Overview
Adds managed auth to the MCP server (documented tool count 10 → 13): three new manage_* tools wired to the Kernel SDK in route.ts.

manage_auth_connections drives profile+domain auth: create (with validation for Kernel vs external credential references), list/get/delete, login (hosted + live view URLs), and submit for MFA fields, MFA option IDs, or SSO button selectors, with optional proxy routing.

manage_credentials exposes list/get, totp_code, and create/update/delete for stored credentials (including values, sso_provider, and totp_secret on write paths).

manage_credential_providers covers external providers (e.g. 1Password): full lifecycle plus list_items and test, including service-account tokens on create/update.

The README lists the new capabilities; the rest of the diff is mostly formatting on existing tools (e.g. computer_action).

^{Reviewed by Cursor Bugbot for commit 0f6eb57. Bugbot is set up for automated code reviews on this repo. Configure here.}

…manage_credential_providers) Closes the largest agent-facing capability gap in the MCP server: setting up an authenticated browser session for a third-party site. Agents can now drive Kernel's managed auth flow end-to-end. - manage_auth_connections: create/list/get/delete connections; start login flows (returns hosted_url + live_view_url); submit MFA codes or SSO selections when a flow is awaiting input. - manage_credentials: read-only (list, get, totp_code). Credentials are created by humans via dashboard/CLI so the agent never sees raw secrets in its context — it references credentials by name. - manage_credential_providers: read-only (list, get) for external providers like 1Password. Same human-creates / agent-consumes pattern.

vercel · 2026-05-24T19:56:52Z

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
mcp	Ready	Preview, Comment	May 26, 2026 1:14am

firetiger-agent · 2026-05-24T20:05:47Z

Monitoring Plan: Add managed-auth and credentials MCP tools

What this PR does: Registers three new MCP tool handlers — manage_auth_connections (full CRUD + login/submit on managed auth connections), manage_credentials (read-only: list, get, TOTP code fetch), and manage_credential_providers (read-only: list, get) — by adding ~450 lines to the MCP server route handler. The remaining diff is formatting-only changes to existing tools.

Intended effect: After deploy, AI agents using the MCP server can invoke these three tools. Successful calls will appear as spans on the already-active Kernel API backend endpoints (/auth/connections/*, /credentials/*, /org/credential_providers). Pre-deploy baseline on those endpoints: 210–983 calls/hr on auth connections, 144–173 calls/hr on credentials — both with 0 errors in the last 24h. Confirmation: first tool invocations produce new spans on those routes with no error status.

Risks:

Tool registration crash — Railway HTTP 5xx rate (all routes); alert if > 15/hr sustained (baseline: 3–8/hr)
Auth connection API errors — /auth/connections trace spans with status.code = 2; alert if any errors appear post-deploy (baseline: 0/hr)
Credentials API errors — /credentials + /org/credential_providers trace spans with status.code = 2; alert if any errors appear post-deploy (baseline: 0/hr)
SDK method missing at runtime — TypeError: Cannot read properties in Railway HTTP error logs; alert on any such log matching manage_auth_connections, manage_credentials, or manage_credential_providers
TOTP misconfiguration — /credentials/{id_or_name}/totp-code returning 5xx; alert on any 5xx on this endpoint

Status updates will be posted automatically on this PR as monitoring progresses.

View agent

Cursor Bugbot caught that we were sending invalid credential payloads when only credential_path or credential_auto was provided without credential_provider. Add upfront validation so the agent gets a clear MCP-level error instead of a generic API rejection.

Cursor Bugbot follow-up: credential_auto=false was treated as a valid provider variant, and credential_path + credential_auto: true were silently allowed together. Only credential_auto: true now counts as a provider mode, and path/auto are enforced as mutually exclusive.

…providers also expose list_items and test on credential providers. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

cursor

Cursor Bugbot has reviewed your changes and found 4 potential issues.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit e945e6e. Configure here.}

cursor · 2026-05-26T01:07:46Z

+    {
+      action: z
+        .enum(["list", "get", "totp_code", "create", "update", "delete"])
+        .describe("Operation to perform."),


Agents can store raw secrets

High Severity

manage_credentials and manage_credential_providers expose create, update, and delete actions that accept passwords, field maps, and provider tokens. The PR states credentials and providers are human-managed only so agents never handle raw secrets; those write paths let an agent persist secrets via MCP.

Additional Locations (1)

src/app/[transport]/route.ts#L2714-L2724

^{Reviewed by Cursor Bugbot for commit e945e6e. Configure here.}

…t empty credential values Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

vercel Bot deployed to Preview May 24, 2026 19:57 View deployment

masnwilliams marked this pull request as ready for review May 24, 2026 20:01

Merge branch 'main' into hypeship/add-managed-auth-tools

c3bd9d6

vercel Bot deployed to Preview May 24, 2026 20:02 View deployment

cursor Bot reviewed May 24, 2026

View reviewed changes

Comment thread src/app/[transport]/route.ts

vercel Bot deployed to Preview May 25, 2026 01:53 View deployment

cursor Bot reviewed May 25, 2026

View reviewed changes

Comment thread src/app/[transport]/route.ts Outdated

masnwilliams closed this May 25, 2026

masnwilliams reopened this May 25, 2026

vercel Bot deployed to Preview May 25, 2026 03:28 View deployment

add create/update/delete to manage_credentials and manage_credential_…

e945e6e

…providers also expose list_items and test on credential providers. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

vercel Bot deployed to Preview May 26, 2026 01:06 View deployment

cursor Bot reviewed May 26, 2026

View reviewed changes

address bugbot: validate auth submit body, check create result, rejec…

0f6eb57

…t empty credential values Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

vercel Bot deployed to Preview May 26, 2026 01:14 View deployment

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

add managed auth tools (manage_auth_connections, manage_credentials, manage_credential_providers)#104

add managed auth tools (manage_auth_connections, manage_credentials, manage_credential_providers)#104
masnwilliams wants to merge 6 commits into
mainfrom
hypeship/add-managed-auth-tools

masnwilliams commented May 24, 2026 •

edited by cursor Bot

Loading

Uh oh!

vercel Bot commented May 24, 2026 •

edited

Loading

Uh oh!

Uh oh!

firetiger-agent Bot commented May 24, 2026

Uh oh!

Uh oh!

cursor Bot left a comment

Uh oh!

cursor Bot May 26, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

masnwilliams commented May 24, 2026 • edited by cursor Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

What's added

Agent flow

Test plan

Tool count

Uh oh!

vercel Bot commented May 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

firetiger-agent Bot commented May 24, 2026

Monitoring Plan: Add managed-auth and credentials MCP tools

Uh oh!

Uh oh!

cursor Bot left a comment

Choose a reason for hiding this comment

Uh oh!

cursor Bot May 26, 2026

Choose a reason for hiding this comment

Agents can store raw secrets

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

masnwilliams commented May 24, 2026 •

edited by cursor Bot

Loading

vercel Bot commented May 24, 2026 •

edited

Loading