Skip to content

DO NOT MERGE: #39551 real-machine evidence (base) - #39806

Closed
chenmingwei23 wants to merge 3 commits into
mainfrom
evidence-39551-base
Closed

chenmingwei23 wants to merge 3 commits into
mainfrom
evidence-39551-base

Conversation

@chenmingwei23

Copy link
Copy Markdown
Contributor

Evidence only for #39551; will be closed.

Evidence only. Spawns the namespace sandbox with a caller-hidden file and
reports whether renaming its parent folder is refused.
@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Nov 20, 2026
@github-actions

github-actions Bot commented Nov 20, 2026 •

Copy link
Copy Markdown
Contributor

UX Review (Opus 5.5) — ⏭️ skipped

Revision dcb2e1c8dc8bbefe695ce5b1c23d37198ba6e064 touches no user-facing surface (no changes under website/ or committed screenshots), so the UX review was skipped. Does not block merge.

@github-actions

github-actions Bot commented Nov 20, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Opus 5.5) — ✅ PASS

Premise-level review of dcb2e1c8dc8bbefe695ce5b1c23d37198ba6e064 — why this exists and whether the shipped surface is the smallest honest version. Updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

First-Principles-Verdict: PASS

Evidence-only PR that the author says will be closed. Its one real change is a strict-xfail test. Every other hunk is #39317, which is already on main.

Notes

What this change ships

Inventory (2 items) — 1 justified

Intent: show on the base commit that #39551's caller-parent rename pins are missing, as real-machine evidence. This is an ADDITION (evidence).

  1. Everything for --use-operator-credentials (flag, preflight probes, risk line, credentials field, docs, decision entry) — justified
  2. A strict-xfail namespace test that renames a parent folder and prints a REALMACHINE-39551 warning line — oversized: an evidence probe that would sit on main

[FIRST-PRINCIPLES-REVIEWED] dcb2e1c

Description read

[DESCRIPTION-READ] aee4d96a8b8c855e7ac5583bc3746990300a4867186ca023717cd03973e93e1b

That sha256 names the PR description this verdict read. Recompute it with .github/scripts/pr-description-capture.sh; if it differs, the description changed after this verdict was formed and any finding drawn from the description is unproven.

@github-actions

github-actions Bot commented Nov 20, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Opus 5.5) — ✅ PASS

Design-level review of dcb2e1c8dc8bbefe695ce5b1c23d37198ba6e064 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

Design-Verdict: PASS

Nothing to check. This is an evidence-only PR that will be closed and only adds a strict-xfail test. It renders nothing new and takes nothing away.

[DESIGN-REVIEWED] dcb2e1c

Description read

[DESCRIPTION-READ] b070445af04352e46021126cd917b5d79d46f41ecac81bebbed36f2d069d3768

That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Nov 20, 2026 •

Copy link
Copy Markdown
Contributor

GPT 6.1 Review — 🔴 changes requested (blocking)

GPT 6.1 found at least one blocking issue that must be resolved before merging dcb2e1c8dc8bbefe695ce5b1c23d37198ba6e064. 1 of 1 blocking finding(s) are security-class and were withheld from adjudication, so the blocking verdict stands.

This comment is updated in place on each push.

BLOCKING -- test/test_sandbox_hidden_leaf_parent_pins.py:171 -- Unconditional xfail makes the sandbox denial failure non-blocking

@pytest.mark.xfail(strict=True, reason="main lacks caller-parent pins; #39551 adds them")
Trigger: The namespace-sandbox CI job runs this test on Linux with namespace support. namespace_argv omits caller-hidden ancestors from its parent pins, so the child renames the parent and fails the EBUSY assertion.
Outcome: Pytest converts that failure into a successful XFAIL, allowing the security regression check to remain green despite the failed denial.
Anchor: a-ratchet-may-only-tighten
Fix: Keep base-only reproduction evidence outside the mergeable suite; retain these assertions without xfail when landing the corresponding fix.
[BLOCK-MERGE] dcb2e1c
[GPT-REVIEWED] dcb2e1c

Adjudication (Opus 5.5) — is blocking on each finding proportionate?

F1 (fenced): the PR adds one test file (git diff --stat HEAD^1 HEAD shows only test/test_sandbox_hidden_leaf_parent_pins.py, 58 insertions) and changes nothing in src/. The rename gap the test shows in sandbox_argv.py:843 namespace_argv already exists on main, so the PR exposes nothing new.
The xfail at test/test_sandbox_hidden_leaf_parent_pins.py:171 is strict=True, so CI goes red with XPASS as soon as #39551 adds the pins, which forces the marker off. What stays is a known, tracked gap, not a security check that was silenced.
The fix is cheap: drop the marker when #39551 lands, or keep the evidence outside the suite. This is an override rationale only; the finding keeps blocking.

[ADJUDICATION] dcb2e1c8dc8bbefe695ce5b1c23d37198ba6e064 total=0 uphold=0 downgrade=0
[GPT-ADJUDICATED] dcb2e1c8dc8bbefe695ce5b1c23d37198ba6e064

[ADJUDICATION-FENCED] dcb2e1c8dc8bbefe695ce5b1c23d37198ba6e064 fenced=1 flagged=1
FLAG F1 test/test_sandbox_hidden_leaf_parent_pins.py:171 -- Test-only diff; the rename gap already exists on main and is tracked by #39551, and strict=True makes CI fail the moment the fix lands, so nothing that currently guards anything is weakened.
[GPT-ADJUDICATED-FENCED] dcb2e1c8dc8bbefe695ce5b1c23d37198ba6e064

🏷️ Fenced finding(s) machine-flagged as likely edge case

The security fence keeps these findings blocking regardless of adjudication; the only clearance path is a human override. The annotate-only pass judged the condition combination each one requires extreme, and pre-drafted the override rationale below. A repository writer must independently verify a rationale before posting it — it is machine-authored, and a wrong override on a security-class finding ships exactly the class the fence exists to stop.

  • F1 test/test_sandbox_hidden_leaf_parent_pins.py:171 — Test-only diff; the rename gap already exists on main and is tracked by fix: guard parent folders of caller-provided hidden paths in the Linux sandbox #39551, and strict=True makes CI fail the moment the fix lands, so nothing that currently guards anything is weakened.

    /ai-review override gpt dcb2e1c8dc8bbefe695ce5b1c23d37198ba6e064 <class>: Test-only diff; the rename gap already exists on main and is tracked by #39551, and strict=True makes CI fail the moment the fix lands, so nothing that currently guards anything is weakened.
    

Replace <class> with the class that records what you did. These findings are real, so for an accepted risk use owner-decision (link the decision comment in the reason) or deferred #<N> (a tracked follow-up issue); false-positive only if you verified the finding is wrong. Not not-applicable: that class files the override as a rule mismatch in the audit.

False positive or not applicable? A repository writer can comment:
/ai-review override gpt dcb2e1c8dc8bbefe695ce5b1c23d37198ba6e064 false-positive: <reason, 20 to 500 characters>

@github-actions

github-actions Bot commented Nov 20, 2026 •

Copy link
Copy Markdown
Contributor

Opus 5.5 Review — ✅ no blocking findings

Reviewed dcb2e1c8dc8bbefe695ce5b1c23d37198ba6e064 — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] dcb2e1c

Verdict parsed from the review's SHA-scoped output markers for commit dcb2e1c8dc8bbefe695ce5b1c23d37198ba6e064.

False positive or not applicable? A repository writer can comment:
/ai-review override fable dcb2e1c8dc8bbefe695ce5b1c23d37198ba6e064 false-positive: <reason, 20 to 500 characters>

Main has no caller-parent pins, so the rename succeeds there; the test is
strict so a pass would show up.
@chenmingwei23

Copy link
Copy Markdown
Contributor Author

Evidence for #39551 recorded in its body; closing (not merging).

@github-actions github-actions Bot removed the readiness: checking Automated validation is still running label Nov 20, 2026
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