Restrict Google Discovery bundle URLs

[RhysSullivan]

What changed

• Canonicalize Google Discovery URLs before fetch, storage, and detection.
• Allow only HTTPS www.googleapis.com/discovery/v1/apis/.../rest and documented {service}.googleapis.com/$discovery/rest?version=... endpoints.
• Reject lookalike hosts, userinfo, non-HTTPS schemes, and unexpected query parameters before making bundle fetches.
• Keep preset audience warnings aligned with canonical Discovery URLs.

Why

The Google bundle flow accepted caller-supplied Discovery URLs and only checked for broad googleapis.com strings, which allowed lookalike hosts and untrusted fetch targets.

Validation

• bun --bun vitest run src/sdk/discovery.test.ts src/sdk/plugin.test.ts src/sdk/product-picker-scopes.test.ts from packages/plugins/google
• bun run --cwd packages/plugins/google typecheck
• git diff --check

Stack

Base: fix/openapi-hosted-origin-guard

Previous: #1104

Next: fix/microsoft-graph-url-allowlist

[cloudflare-workers-and-pages]

Deploying with    Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status	Name	Latest Commit	Preview URL	Updated (UTC)
✅ Deployment successful! View logs	executor-marketing	f275c2c	Commit Preview URL Branch Preview URL	Jun 23 2026, 06:13 PM

[cloudflare-workers-and-pages]

Deploying with    Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status	Name	Latest Commit	Updated (UTC)
✅ Deployment successful! View logs	executor-cloud	f275c2c	Jun 23 2026, 06:16 PM

[github-actions]

Cloudflare preview

Torn down — the PR is closed.

[pkg-pr-new]

Open in StackBlitz

@executor-js/cli

npm i https://pkg.pr.new/@executor-js/cli@1105

@executor-js/config

npm i https://pkg.pr.new/@executor-js/config@1105

@executor-js/execution

npm i https://pkg.pr.new/@executor-js/execution@1105

@executor-js/sdk

npm i https://pkg.pr.new/@executor-js/sdk@1105

@executor-js/codemode-core

npm i https://pkg.pr.new/@executor-js/codemode-core@1105

@executor-js/runtime-quickjs

npm i https://pkg.pr.new/@executor-js/runtime-quickjs@1105

@executor-js/plugin-file-secrets

npm i https://pkg.pr.new/@executor-js/plugin-file-secrets@1105

@executor-js/plugin-graphql

npm i https://pkg.pr.new/@executor-js/plugin-graphql@1105

@executor-js/plugin-keychain

npm i https://pkg.pr.new/@executor-js/plugin-keychain@1105

@executor-js/plugin-mcp

npm i https://pkg.pr.new/@executor-js/plugin-mcp@1105

@executor-js/plugin-onepassword

npm i https://pkg.pr.new/@executor-js/plugin-onepassword@1105

@executor-js/plugin-openapi

npm i https://pkg.pr.new/@executor-js/plugin-openapi@1105

executor

npm i https://pkg.pr.new/executor@1105

commit: 54782a5

[greptile-apps]

Greptile Summary

This PR hardens the Google Discovery bundle flow by replacing broad googleapis.com substring checks with a strict two-branch allowlist: www.googleapis.com/discovery/v1/apis/{service}/{version}/rest and {service}.googleapis.com/$discovery/rest?version=…. Both branches canonicalize to the same www.googleapis.com form, so all three entry points (fetchGoogleDiscoveryDocument, uniqueUrls, detect) now operate exclusively on validated canonical URLs.

• normalizeGoogleDiscoveryUrl (new public export) rejects lookalike hosts, non-HTTPS schemes, userinfo, hash fragments, unexpected query parameters, and any pathname that doesn't match a known Discovery pattern before any network call is made.
• uniqueUrls now deduplicates by canonical URL rather than raw string, so chat.googleapis.com/$discovery/rest?version=v1 and its www.googleapis.com equivalent correctly collapse to one entry.
• presets.ts wires normalizeGoogleDiscoveryUrl into the preset-URL normalization so $discovery/rest-style preset URLs (forms, keep, admin, cloudresourcemanager) resolve to canonical form for audience-warning lookups.

Confidence Score: 5/5

Safe to merge — the change adds defence-in-depth URL allowlisting with no regressions to the existing fetch or detect flows.

All three entry points now validate and canonicalize Discovery URLs before any network activity. The two-branch normalization is idempotent (canonical URLs round-trip cleanly), preset lookups and audience warnings are correctly updated, and the new tests exercise both the happy path and lookalike-host rejection. No pre-existing flows are broken, and the new guard is consistently applied.

No files require special attention; discovery.ts carries the most logic and has solid test coverage.

Important Files Changed

Filename	Overview
packages/plugins/google/src/sdk/discovery.ts	Core URL validation logic: exports normalizeGoogleDiscoveryUrl and isGoogleDiscoveryUrl; adds strict allowlist for HTTPS googleapis.com endpoints; preflight check in fetchGoogleDiscoveryDocument blocks fetches to non-canonical hosts.
packages/plugins/google/src/sdk/plugin.ts	Switches uniqueUrls from a trim/length filter to normalizeGoogleDiscoveryUrl; replaces isGoogleDiscoveryUrl with normalizeGoogleDiscoveryUrl in detect so the canonical URL is used throughout the add/detect flows.
packages/plugins/google/src/sdk/presets.ts	normalizeGooglePresetUrl now tries normalizeGoogleDiscoveryUrl first, ensuring preset URLs stored in $discovery/rest format (forms, keep, admin, cloudresourcemanager) are canonicalized before audience-warning lookups.
packages/plugins/google/src/sdk/discovery.test.ts	Adds unit tests for normalizeGoogleDiscoveryUrl (trailing-slash stripping, service subdomain → canonical conversion) and isGoogleDiscoveryUrl (lookalike host, HTTP, extra query params, userinfo rejection).
packages/plugins/google/src/sdk/plugin.test.ts	Adds integration test that verifies no HTTP requests are made when all supplied URLs are lookalike hosts; confirms the failure exits cleanly without touching the mocked HttpClient.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[Caller-supplied URL] --> B[normalizeGoogleDiscoveryUrl]
    B --> C{protocol === https:\nno userinfo/hash?}
    C -->|No| D[return null]
    C -->|Yes| E{host === www.googleapis.com?}
    E -->|Yes| F{any search params?}
    F -->|Yes| D
    F -->|No| G{pathname matches\n/discovery/v1/apis/service/version/rest?}
    G -->|No| D
    G -->|Yes| H[canonical www.googleapis.com URL]
    E -->|No| I{serviceFromGoogleApisHost\nhost.endsWith .googleapis.com\nno subdots, valid chars}
    I -->|null| D
    I -->|service| J{pathname /$discovery/rest\nonly version= param\nversion passes RE?}
    J -->|No| D
    J -->|Yes| H
    H --> K[uniqueUrls dedup by canonical form]
    H --> L[fetchGoogleDiscoveryDocument\npreflight re-validates]
    H --> M[detect: store canonical endpoint]
    H --> N[presets: audience-warning lookup]

Loading

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
flowchart TD
    A[Caller-supplied URL] --> B[normalizeGoogleDiscoveryUrl]
    B --> C{protocol === https:\nno userinfo/hash?}
    C -->|No| D[return null]
    C -->|Yes| E{host === www.googleapis.com?}
    E -->|Yes| F{any search params?}
    F -->|Yes| D
    F -->|No| G{pathname matches\n/discovery/v1/apis/service/version/rest?}
    G -->|No| D
    G -->|Yes| H[canonical www.googleapis.com URL]
    E -->|No| I{serviceFromGoogleApisHost\nhost.endsWith .googleapis.com\nno subdots, valid chars}
    I -->|null| D
    I -->|service| J{pathname /$discovery/rest\nonly version= param\nversion passes RE?}
    J -->|No| D
    J -->|Yes| H
    H --> K[uniqueUrls dedup by canonical form]
    H --> L[fetchGoogleDiscoveryDocument\npreflight re-validates]
    H --> M[detect: store canonical endpoint]
    H --> N[presets: audience-warning lookup]

Loading

_{Reviews (2): Last reviewed commit: "Restrict Google Discovery bundle URLs" | Re-trigger Greptile}

[RhysSullivan]

Superseded by batch merge #1106.