test(proxy): harden transcript-proxy guard against two silent bypasses (hardens #1129) by groupthinking · Pull Request #1131 · groupthinking/EventRelay

groupthinking · 2026-07-31T07:27:32Z

Canonical issue

Closes #1087 (via the same fix as #1129 — see Agent handoff below).

Outcome

Strengthens the AST regression guard introduced by #1129 so it can no longer
pass a YouTubeTranscriptApi(...) construction that would still egress outside
the centralized outbound proxy. This branch is fix/centralized-transcript-proxy-1087's
HEAD (5659f2b) plus one test-only commit; the production fix is unchanged.

Scope

Included: two test-only hardenings in tests/unit/test_transcript_proxy_coverage.py:
1. _has_required_keyword no longer treats **kwargs-only calls as compliant —
  YouTubeTranscriptApi(**opts) reads as opaque but egresses directly whenever
  opts omits proxy_config. It now requires an explicit proxy_config= keyword.
2. test_module_resolves_proxy_config_from_the_canonical_helper now inspects each
  construction's proxy_config AST value and requires it to be a call to
  get_transcript_proxy_config(). Previously a substring/import check let
  proxy_config=<foreign config> or proxy_config=object() pass.
Explicitly excluded: no production/source changes; no reformatting of unrelated lines.

Risk

Risk level: low
Failure mode: test-only. The two tightenings add zero false positives — all 14
live constructions in src/ pass proxy_config=get_transcript_proxy_config() explicitly.
Rollback: revert the single commit 1ba9cd4.

Verification

All at head 1ba9cd4666c1339c7c3001cafba21f8180639fb4.

Focused tests — pytest tests/unit/test_transcript_proxy_coverage.py → 21 passed.
Mutation-verified the hardenings are not vacuous:
- a YouTubeTranscriptApi(**opts) bypass now fails test_every_transcript_client_passes_proxy_config (previously passed);
- a proxy_config=object() construction now fails test_module_resolves_proxy_config_from_the_canonical_helper (previously passed).
ruff check clean; black --check clean on the changed file.
Required CI — will run on this PR.
Review threads resolved — the two Copilot findings on fix: route every transcript client through the centralized proxy (#1087) #1129 are addressed by this diff.

Production evidence

Not applicable — test-only change. The production behaviour and its evidence are
carried by #1129 (Vercel green, CodeRabbit approved).

Agent handoff

One canonical issue is linked (Migrate YouTube transcript client to v1.x with proxy support #1087)
No competing PR implements the same issue — ⚠️ fix: route every transcript client through the centralized proxy (#1087) #1129 does. This branch is a
strict superset of fix: route every transcript client through the centralized proxy (#1087) #1129 (its HEAD + one test hardening). They must not both merge.
Recommended: merge this one and close fix: route every transcript client through the centralized proxy (#1087) #1129, or cherry-pick commit 1ba9cd4 onto
fix: route every transcript client through the centralized proxy (#1087) #1129 and close this. Human's call.
Acceptance criteria satisfied (14/14 constructions proxied; guard hardened)
Required checks pass on current head (pending CI)
Human decision requested only for the merge/dedup choice above (irreversible: merge to protected main)

Generated by Claude Code

youtube-transcript-api >=1.0 only honours a proxy when a ``proxy_config`` is passed to the constructor. Six call sites across four modules built a bare ``YouTubeTranscriptApi()``, so they egressed from the host's own IP and bypassed WEBSHARE_PROXY_URL entirely — including ``_extract_transcript_with_rotation``, whose name implies the opposite. - src/integration/youtube_api.py:99 - src/agents/process_video_with_mcp.py:222,233 - src/agents/interactive_metadata_extractor.py:87 - src/mcp/mcp_video_processor.py:694,718 Each now resolves its config from the canonical ``youtube_extension.utils.proxy.get_transcript_proxy_config`` helper, guarded by the import fallback this repo already uses in gemini_video_master_agent.py so modules stay importable outside the package. The helper returns None when no proxy is configured, so direct connections keep working unchanged. Adds tests/unit/test_transcript_proxy_coverage.py: an AST walk over src/ and shared/ that fails if any construction omits proxy_config, plus a canary that fails if the scan ever matches nothing (the vacuity mode that has bitten this repo repeatedly). Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

…roxy Review on #1129 found the guarded-import fallback introduced by this branch was itself a silent bypass: when a module is executed by path (the documented CLI entry point) youtube_extension is not importable, so the stub returned None and proxy_config=get_transcript_proxy_config() egressed directly from the host IP -- the exact failure this issue exists to prevent. Replace the stub in all four standalone-executable modules with a sys.path bootstrap to the repository's src directory, falling back to a hard ImportError. The proxy is now either honoured or the module refuses to load; it is never silently skipped. Harden the guard suite against three ways it could pass while protecting less than it claims: - a missing configured source root was skipped silently - SOURCE_ROOTS could be narrowed by hand with every other test still green; the required roots are now rediscovered from the tree - referencing the helper was satisfied by the call site alone, so a broken import passed; every import from the proxy module must now bind the helper All 9 mutations kill their tests, including one that revealed a newly added test was itself vacuous (it asserted parents[1] from a test-side constant rather than the index the source actually uses). Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com>

Closes the two residual holes the automated review flagged on the guard added by #1129 — both were ways a construction could pass the guard while still egressing outside the centralized proxy: 1. `**kwargs`-only calls were treated as compliant. `YouTubeTranscriptApi( **opts)` reads as opaque but egresses directly whenever `opts` omits `proxy_config` — the exact silent bypass this guard exists to catch. `_has_required_keyword` now requires an explicit `proxy_config=` keyword. No real call site uses `**kwargs`, so this adds zero false positives. 2. `test_module_resolves_proxy_config_from_the_canonical_helper` only checked that the helper name appeared in the source and was imported; a call passing `proxy_config=<foreign config>` or `proxy_config=object()` still passed. It now inspects each construction's `proxy_config` AST value and requires it to be a call to `get_transcript_proxy_config()`, subsuming the old literal-None check. Both tightenings are mutation-verified: a `**kwargs` bypass and a `proxy_config=object()` construction each now fail the guard where they previously passed. All 21 tests green against the real (14/14 compliant) call sites. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NaRe39Fsrq2tv8vpn2aRdp

vercel · 2026-07-31T07:27:39Z

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

Project	Deployment	Actions	Updated (UTC)
v0-uvai	Ready	Preview, v0	Jul 31, 2026 7:28am

github-actions · 2026-07-31T07:27:45Z

🔍 PR Validation

⚠️ Large PR detected (559 lines changed)

github-actions · 2026-07-31T07:27:51Z

Dependency Review

✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.

Snapshot Warnings

⚠️: No snapshots were found for the head SHA 1ba9cd4.

Ensure that dependencies are being submitted on PR branches and consider enabling retry-on-snapshot-warnings. See the documentation for more information and troubleshooting advice.

Scanned Files

None

github-actions · 2026-07-31T07:28:03Z

Agent Completion Truth Gate: BLOCKED

Reasons: invalid_payload

Machine-readable verdict

{
  "details": {
    "collection_errors": [
      "incomplete_linked_issue_contract",
      "linked_issue_not_agent_task",
      "missing_intent_snapshot",
      "missing_agent_run_id",
      "missing_agent_login"
    ],
    "invalid_fields": [
      "policy.agent_login",
      "policy.run_id"
    ]
  },
  "reasons": [
    "invalid_payload"
  ],
  "verdict": "blocked"
}

Workflow evidence

groupthinking · 2026-07-31T07:29:56Z

Blocker: Agent Completion Truth Gate — and why I'm not forcing past it

agent-completion/truth-gate/pr-1131 is failing (invalid_payload → policy.agent_login, policy.run_id; earlier missing_trusted_publication). This is expected, not a code defect:

The change is on a claude/… branch, so the gate classifies it as agent-authored and requires a filled agent-lock manifest backed by a trusted publication (a real agent run_id + published head_sha event). The canonical PR this supersedes, fix: route every transcript client through the centralized proxy (#1087) #1129, passed the gate as not_applicable because its fix/… branch isn't agent-classified.
I'm an unattended session with no trusted run_id to publish. I will not fabricate agent_login/run_id to green a truth gate — that would defeat exactly the control it exists to enforce. So per drive-to-green, I'm reporting the blocker rather than gaming it.

The change itself is sound and test-only

Two hardenings to the regression guard from #1129, closing the two residual holes the Copilot review flagged (commit 1ba9cd4):

_has_required_keyword no longer accepts **kwargs-only calls — an explicit proxy_config= is required.
test_module_resolves_proxy_config_from_the_canonical_helper now inspects each construction's proxy_config AST value and requires a call to get_transcript_proxy_config() (previously a substring/import check let proxy_config=object() through).

Both are mutation-verified (a **kwargs bypass and a proxy_config=object() construction each now fail the guard where they previously passed); pytest → 21 passed; ruff/black clean.

Recommended human action

This branch is a strict superset of #1129 (its HEAD 5659f2b + one test commit). Don't merge both. Cleanest path:

Cherry-pick 1ba9cd4 onto fix: route every transcript client through the centralized proxy (#1087) #1129 and merge fix: route every transcript client through the centralized proxy (#1087) #1129 (already green + CodeRabbit-approved), then close this PR — the guard hardening lands on a branch the truth gate accepts, and the duplicate 540-line diff disappears from the queue.

Merging to protected main is a human decision; I've deliberately not auto-merged anything.

Generated by Claude Code

coderabbitai · 2026-07-31T07:39:13Z

Important

Review skipped

Auto reviews are limited based on label configuration.

🏷️ Required labels (at least one) (1)

[‘architecture-gap’, ‘bug’, ‘ci-cd’, ‘ci/cd’, ‘copilot-rabbit’, ‘documentation’, ‘duplicate’, ‘enhancement’, ‘frontend’, ‘github_actions’, ‘good first issue’, ‘help wanted’, ‘high-priority’, ‘invalid’, ‘javascript’, ‘ml-model’, ‘needs-triage’, ‘pipeline-critical’, ‘placeholder-code’, ‘priority:high’, ‘python’, ‘python:uv’, ‘question’, ‘styling’, ‘tests’, ‘v0’]

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository YAML (base), Repository UI (inherited), Organization UI (inherited)

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 43f652a4-a06a-4a49-aa13-55fc6a07402a

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

🔍 Trigger review

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands.}

groupthinking · 2026-07-31T08:09:18Z

Contained as a duplicate of canonical #1129. The exact two-part test-only hardening from 1ba9cd4 was transferred to canonical branch fix/centralized-transcript-proxy-1087 at commit 49e887c; the resulting test blob is identical (da6871e5ed13f06afac08b71da7f6089d1936c62). No production changes from this duplicate were needed. The branch is preserved.

groupthinking and others added 3 commits July 31, 2026 01:08

github-actions Bot added the python label Jul 31, 2026

vercel Bot deployed to Preview July 31, 2026 07:28 View deployment

groupthinking added the duplicate This issue or pull request already exists label Jul 31, 2026 — with ChatGPT Codex Connector

groupthinking closed this Jul 31, 2026

This was referenced Jul 31, 2026

fix: route every transcript client through the centralized proxy (#1087) #1129

Closed

Migrate YouTube transcript client to v1.x with proxy support #1087

Open

[PROGRAM][OWNER:Codex] Autonomous EventRelay delivery control plane #898

Open

Navigation Menu

test(proxy): harden transcript-proxy guard against two silent bypasses (hardens #1129) - #1131

test(proxy): harden transcript-proxy guard against two silent bypasses (hardens #1129)#1131
groupthinking wants to merge 3 commits into
mainfrom
claude/determined-maxwell-4k7sox

groupthinking commented Jul 31, 2026 •

edited

Loading

Uh oh!

vercel Bot commented Jul 31, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Jul 31, 2026

Uh oh!

github-actions Bot commented Jul 31, 2026

Uh oh!

github-actions Bot commented Jul 31, 2026 •

edited

Loading

Uh oh!

groupthinking commented Jul 31, 2026

Uh oh!

coderabbitai Bot commented Jul 31, 2026 •

edited

Loading

Review skipped

Uh oh!

groupthinking commented Jul 31, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

groupthinking commented Jul 31, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Canonical issue

Outcome

Scope

Risk

Verification

Production evidence

Agent handoff

Uh oh!

vercel Bot commented Jul 31, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Jul 31, 2026

🔍 PR Validation

Uh oh!

github-actions Bot commented Jul 31, 2026

Dependency Review

Snapshot Warnings

Scanned Files

Uh oh!

github-actions Bot commented Jul 31, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Agent Completion Truth Gate: BLOCKED

Uh oh!

groupthinking commented Jul 31, 2026

Blocker: Agent Completion Truth Gate — and why I'm not forcing past it

The change itself is sound and test-only

Recommended human action

Uh oh!

coderabbitai Bot commented Jul 31, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Review skipped

Uh oh!

groupthinking commented Jul 31, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

groupthinking commented Jul 31, 2026 •

edited

Loading

vercel Bot commented Jul 31, 2026 •

edited

Loading

github-actions Bot commented Jul 31, 2026 •

edited

Loading

coderabbitai Bot commented Jul 31, 2026 •

edited

Loading