Enforce configured targets in safe-output handlers by Copilot · Pull Request #62447 · github/gh-aw

Copilot · 2026-09-21T19:14:51Z

Several safe-output handlers selected model-provided IDs before applying their configured target, allowing operations to escape the default triggering item or a fixed numeric target. link-sub-issue also dropped its parsed target during handler configuration.

Changes

Target enforcement
- Apply target before model-provided IDs across merge, comment, close, review, assignment, milestone, and issue-field handlers.
- Preserve explicit and temporary IDs only for target: "*".
- Keep omitted targets equivalent to target: triggering.
Boundary validation
- Validate hide-comment targets against the comment’s parent item and repository.
- Enforce fixed and triggering targets for pull-request review threads, including legacy same-repository mode.
- Reject review-thread operations without a triggering PR when using the default target.
Sub-issue configuration
- Propagate link-sub-issue.target into runtime configuration.
- Apply the target to the parent issue while retaining the model-provided sub-issue.
- Enforce same-repository parent/sub-issue relationships.
Compiler validation
- Register target validation for unassign-from-user, set-issue-type, set-issue-field, and resolve-pull-request-review-thread.

safe-outputs:
  close-issue:
    target: triggering # model-provided issue numbers are ignored
  assign-milestone:
    target: "123"      # always targets issue #123
  add-reviewer:
    target: "*"        # model-provided PR numbers are allowed

pr-sous-chef
Run: https://github.com/github/gh-aw/actions/runs/35656487865

Generated by 👨‍🍳 PR Sous Chef · pi · gpt54 · 27.7 AIC · ⊞ 9.6K · ◷
_{Comment /souschef to run again}

Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com>

pelikhan · 2026-09-21T19:56:23Z

@copilot update safe-outputs specification and specification enforcing scripts

github-actions · 2026-09-21T20:59:23Z

Great work on this security boundary hardening! 🎯 The safe-output target enforcement refactor looks solid — consistent application of target validation across merge, comment, close, review, assignment, milestone, and issue-field handlers, plus rigorous boundary checks for cross-repository and review-thread operations.

The changes are well-focused (35 files covering a single concern), comprehensive, and include test coverage. This is ready for review by the maintainers.

Warning

Firewall blocked 1 domain

The following domain was blocked by the firewall during workflow execution:

github.com

To allow these domains, add them to the network.allowed list in your workflow frontmatter:

network:
  allowed:
    - defaults
    - "github.com"

See Network Configuration for more information.

Generated by ✅ Contribution Check · copilot · auto · 50.1 AIC · ⌖ 13.5 AIC · ⊞ 9.2K · ◷

Copilot

Copilot review overview

🟡 Changes recommended

Target enforcement still has parent-type, forwarded-context, effective sub-issue, and PR-assignment correctness gaps.

Get a fresh assessment by requesting another Copilot review.

Review effort: Balanced
Findings: 1 · 4

Open (5)

Match triggering target by parent kind, not just repository and ID · New
Skip ignored ID exclusivity checks for fixed and triggering targets · New
Validate effective target parent and sub-issue pair · New
Resolve forwarded invocation context before triggering PR validation · New
Allow unassign-from-user to target pull requests · New

What changed in this PR

Enforces configured safe-output targets before model-provided IDs and strengthens comment, review-thread, and sub-issue boundaries.

Changes:

Applies target resolution consistently across issue and pull-request handlers.
Adds parent/repository validation for comments, review threads, and sub-issues.
Expands compiler validation and regression coverage.

File	Description
`pkg/workflow/safe_outputs_validation.go`	Expands target validation.
`pkg/workflow/safe_outputs_target_validation_test.go`	Tests added validators.
`pkg/workflow/safe_outputs_handler_registry_issues.go`	Propagates sub-issue target.
`pkg/workflow/link_sub_issue_handler_config_test.go`	Tests target propagation.
`actions/setup/js/unassign_from_user.test.cjs`	Updates target tests.
`actions/setup/js/unassign_from_user.cjs`	Enforces assignment target.
`actions/setup/js/set_issue_type.test.cjs`	Updates target scenarios.
`actions/setup/js/set_issue_type.cjs`	Enforces issue-type target.
`actions/setup/js/set_issue_field.test.cjs`	Updates target scenarios.
`actions/setup/js/set_issue_field.cjs`	Enforces issue-field target.
`actions/setup/js/safe_output_helpers.cjs`	Clarifies target API documentation.
`actions/setup/js/resolve_pr_review_thread.test.cjs`	Tests thread boundaries.
`actions/setup/js/resolve_pr_review_thread.cjs`	Enforces review-thread targets.
`actions/setup/js/merge_pull_request.test.cjs`	Updates merge target setup.
`actions/setup/js/merge_pull_request.cjs`	Enforces merge target.
`actions/setup/js/mark_pull_request_as_ready_for_review.test.cjs`	Updates ready-for-review tests.
`actions/setup/js/mark_pull_request_as_ready_for_review.cjs`	Enforces PR target.
`actions/setup/js/link_sub_issue.test.cjs`	Tests sub-issue targeting.
`actions/setup/js/link_sub_issue.cjs`	Targets parent issue.
`actions/setup/js/hide_comment.test.cjs`	Tests parent lookup behavior.
`actions/setup/js/hide_comment.cjs`	Validates comment parent.
`actions/setup/js/close_pull_request.test.cjs`	Updates close-PR tests.
`actions/setup/js/close_pull_request.cjs`	Enforces close-PR target.
`actions/setup/js/close_issue.test.cjs`	Updates close-issue tests.
`actions/setup/js/close_issue.cjs`	Enforces close-issue target.
`actions/setup/js/assign_to_user.test.cjs`	Updates assignment tests.
`actions/setup/js/assign_to_user.cjs`	Enforces assignee target.
`actions/setup/js/assign_to_agent.test.cjs`	Updates agent target tests.
`actions/setup/js/assign_to_agent.cjs`	Enforces agent target.
`actions/setup/js/assign_milestone.test.cjs`	Updates milestone tests.
`actions/setup/js/assign_milestone.cjs`	Enforces milestone target.
`actions/setup/js/add_reviewer.test.cjs`	Updates reviewer tests.
`actions/setup/js/add_reviewer.cjs`	Enforces reviewer target.
`actions/setup/js/add_comment.test.cjs`	Updates comment target tests.
`actions/setup/js/add_comment.cjs`	Enforces comment target.

💡 Add a code-review agent skill for context-aware, tailored reviews. Learn more in the docs.

Copilot · 2026-09-21T21:55:49Z

+      let expectedNumber;
+      if (targetConfig === "triggering") {
+        const invocationContext = resolveInvocationContext(context);
+        expectedNumber = invocationContext.eventPayload?.issue?.number ?? invocationContext.eventPayload?.pull_request?.number ?? invocationContext.eventPayload?.discussion?.number;


Fixed in de83120: resolveCommentNodeId() now tracks the resolved parent's kind ("issue" or "discussion") alongside its number, and the target: "triggering" check rejects mismatches between the two independent numbering sequences.

Copilot · 2026-09-21T21:55:54Z

    // Defer if issue_number is a temporary ID that hasn't been resolved yet
    // Strip leading '#' so both 'aw_abc1' and '#aw_abc1' (canonical validator form) are handled
-    if (message.issue_number != null) {
+    if (targetConfig === "*" && message.issue_number != null) {


Fixed in de83120: the issue_number/pull_number mutual-exclusivity check now only runs when targetConfig === "*".

Copilot · 2026-09-21T21:55:59Z

    const subIssueNumber = subResolved.resolved?.number;
+    const parentTarget = resolveTarget({
+      targetConfig,
+      item: targetItem,
+      context,
+      itemType: HANDLER_TYPE,
+      supportsIssue: true,
+    });
+    if (!parentTarget.success) {
+      return {
+        parent_issue_number: item.parent_issue_number,
+        sub_issue_number: item.sub_issue_number,
+        success: false,
+        error: parentTarget.error,
+      };
+    }
+    const parentIssueNumber = parentTarget.number;


Fixed in de83120: added a post-target-resolution check that rejects when the effective parent/sub-issue numbers are equal, instead of only validating the raw model-provided parent.

Copilot · 2026-09-21T21:56:04Z

+      if (resolveTarget === "triggering") {
+        if (!triggeringPRNumber) {
+          core.warning("Cannot resolve review thread: not running in a pull request context");
          return {
            success: false,
-            error: repoValidation.error,
+            error: "Cannot resolve review threads outside of a pull request context",


Fixed in de83120: triggeringPRNumber is now derived from resolveInvocationContext(context).eventPayload so forwarded invocations (workflow_dispatch/aw_context, repository_dispatch) are correctly recognized.

Copilot · 2026-09-21T21:56:09Z

+        item: unassignItem,
+        context,
+        itemType: HANDLER_TYPE,
+        supportsIssue: true,


Fixed in de83120: switched to the helper's both-issue-and-PR mode (supportsPR: true) so the default target continues to work for triggering PRs.

gh-aw-bot · 2026-09-21T21:26:26Z

pr-sous-chef
@copilot This PR still needs forward progress before a maintainer can merge it.

Please address the open review follow-ups below, refresh the branch if needed, and then run the pr-finisher skill.

Open review follow-ups (newest first):

Copilot on actions/setup/js/unassign_from_user.cjs: Enforce configured targets in safe-output handlers #62447 (comment)
Copilot on actions/setup/js/resolve_pr_review_thread.cjs: Enforce configured targets in safe-output handlers #62447 (comment)
Copilot on actions/setup/js/link_sub_issue.cjs: Enforce configured targets in safe-output handlers #62447 (comment)
Copilot on actions/setup/js/assign_to_agent.cjs: Enforce configured targets in safe-output handlers #62447 (comment)
Copilot on actions/setup/js/hide_comment.cjs: Enforce configured targets in safe-output handlers #62447 (comment)

I also requested a branch refresh for this PR.
Run: https://github.com/github/gh-aw/actions/runs/35656487865

Generated by 👨‍🍳 PR Sous Chef · pi · gpt54 · 27.7 AIC · ⊞ 9.6K · ◷
_{Comment /souschef to run again}

Copilot · 2026-09-21T22:34:51Z

+      let expectedNumber;
+      if (targetConfig === "triggering") {
+        const invocationContext = resolveInvocationContext(context);
+        expectedNumber = invocationContext.eventPayload?.issue?.number ?? invocationContext.eventPayload?.pull_request?.number ?? invocationContext.eventPayload?.discussion?.number;
+        if (!expectedNumber) {
+          return {
+            success: false,
+            error: 'Target is "triggering" but not running in issue, pull request, or discussion context',
+          };
+        }
+      } else if (targetConfig !== "*") {
+        expectedNumber = Number(targetConfig);
+        if (!Number.isInteger(expectedNumber) || expectedNumber <= 0) {
+          return {
+            success: false,
+            error: `Invalid target configuration: ${targetConfig}`,
+          };
+        }
+      }
+      if (expectedNumber && resolvedComment.itemNumber !== expectedNumber) {
+        return {
+          success: false,
+          error: `Comment belongs to item #${resolvedComment.itemNumber}, but target is #${expectedNumber}`,
+        };
+      }


Addressed in de83120 and reinforced by e5380ea: hide_comment now validates the resolved comment parent kind as well as repository/number, and the shared target regression suite/conformance checks cover configured-target enforcement.

…get Authorization sections Co-authored-by: gh-aw-bot <259018956+gh-aw-bot@users.noreply.github.com>

pelikhan · 2026-09-21T21:57:33Z

@copilot update safe-outputs specification and specification enforcing scripts

Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com>

Copilot · 2026-09-21T22:10:50Z

@copilot update safe-outputs specification and specification enforcing scripts

Updated the safe-output specification and added enforcement tests for set_issue_type and set_issue_field in 7664e4e.

@copilot update safe-outputs specification and specification enforcing scripts

Updated the safe-output specification and added enforcement tests for set_issue_type and set_issue_field in 7664e4e.

pr-sous-chef
@copilot This PR still needs forward progress before a maintainer can merge it.
...

Ran the PR-finisher pass. Local validation is green; existing Copilot review threads already have substantive responses but remain unresolved, and the current GitHub Advanced Security check is still in progress.

pelikhan · 2026-09-21T22:18:13Z

@copilot ensure systematic test suite and fail safe runtime assertion to prevent regression

Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com>

Copilot · 2026-09-21T22:34:51Z

@copilot ensure systematic test suite and fail safe runtime assertion to prevent regression

Implemented in e5380ea/700589a: added systematic safe-output target authorization conformance checks, updated the safe-outputs spec to require target regression testing, and added a fail-safe runtime target authorization invariant with shared JS regression coverage. Local final validation passed (make agent-report-progress).

github-actions · 2026-09-23T04:52:02Z

🎉 This pull request is included in a new release.

Release: v0.89.20

Enforce configured targets in safe output handlers

c011c92

Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com>

Copilot AI assigned Copilot and pelikhan Sep 21, 2026

Copilot AI changed the title ~~Enforce configured targets in safe output handlers~~ Enforce configured targets in safe-output handlers Sep 21, 2026

Copilot created this pull request from a session on behalf of pelikhan September 21, 2026 19:15 View session

Copilot finished work on behalf of pelikhan September 21, 2026 19:15

Copilot AI requested a review from pelikhan September 21, 2026 19:15

Copilot started work on behalf of pelikhan September 21, 2026 19:57 View session

Copilot finished work on behalf of pelikhan September 21, 2026 20:02

github-actions Bot mentioned this pull request Sep 21, 2026

[Contribution Check Report] Contribution Check — 2026-09-21 #62280

Closed

pelikhan marked this pull request as ready for review September 21, 2026 21:03

Copilot AI balanced review requested due to automatic review settings September 21, 2026 21:03

Copilot started reviewing on behalf of pelikhan September 21, 2026 21:03 View session

Copilot AI reviewed Sep 21, 2026

View reviewed changes

Merge branch 'main' into copilot/ensure-target-triggering-enforcement

b16808f

github-actions Bot mentioned this pull request Sep 21, 2026

[pr-sous-chef] PR Sous Chef run report #62242

Closed

github-advanced-security AI found potential problems Sep 21, 2026

View reviewed changes

Copilot started work on behalf of gh-aw-bot September 21, 2026 21:32 View session

Address review feedback: fix target-enforcement gaps and add spec Tar…

de83120

…get Authorization sections Co-authored-by: gh-aw-bot <259018956+gh-aw-bot@users.noreply.github.com>

Copilot finished work on behalf of gh-aw-bot September 21, 2026 21:57

Copilot AI requested a review from gh-aw-bot September 21, 2026 21:57

Copilot started work on behalf of pelikhan September 21, 2026 22:02 View session

Document safe-output target authorization

7664e4e

Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com>

Copilot finished work on behalf of pelikhan September 21, 2026 22:11

Copilot started work on behalf of pelikhan September 21, 2026 22:19 View session

Copilot AI and others added 2 commits September 21, 2026 22:31

Harden safe-output target regression coverage

e5380ea

Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com>

Fix safe-output target invariant type guard

700589a

Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com>

Copilot finished work on behalf of pelikhan September 21, 2026 22:35

pelikhan merged commit abc510c into main Sep 21, 2026
45 checks passed

pelikhan deleted the copilot/ensure-target-triggering-enforcement branch September 21, 2026 23:18

This was referenced Sep 22, 2026

[instructions] Sync instruction files with release v0.89.17 #62592

Merged

Suggested test: copilot/close-issue-fixed-target-enforcement githubnext/gh-aw-test#17292

Open

Conversation

Copilot AI commented Sep 21, 2026 • edited by github-actions Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Changes

Uh oh!

pelikhan commented Sep 21, 2026

Uh oh!

github-actions Bot commented Sep 21, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot review overview

🟡 Changes recommended

Uh oh!

Copilot AI Sep 21, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Sep 21, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Sep 21, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Sep 21, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Sep 21, 2026

Choose a reason for hiding this comment

Uh oh!

gh-aw-bot commented Sep 21, 2026

Uh oh!

Copilot AI Sep 21, 2026

Choose a reason for hiding this comment

Uh oh!

pelikhan commented Sep 21, 2026

Uh oh!

Copilot AI commented Sep 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pelikhan commented Sep 21, 2026

Uh oh!

Copilot AI commented Sep 21, 2026

Uh oh!

Uh oh!

github-actions Bot commented Sep 23, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

Copilot AI commented Sep 21, 2026 •

edited by github-actions Bot

Loading

Copilot AI commented Sep 21, 2026 •

edited

Loading