Reviewer-assignment failure hid a successfully created pull request from the tool response. Entry and authority. The agent can create a PR but requests an invalid or unassignable reviewer. Observed mechanism and consequence. The repository issue shows PR creation succeeded before a 422 reviewer error made the tool return isError without the new PR ID,…
Inspect the ClaimsCase · DiggingBeagle record
GitHub MCP pull-request creation returned failure after committing a PR
Reviewer-assignment failure hid a successfully created pull request from the tool response.
Public disclosure: Nov 9, 2026
Scope
The agent can create a PR but requests an invalid or unassignable reviewer. The repository issue shows PR creation succeeded before a 422 reviewer error made the tool return isError without the new PR ID, complicating safe retries.
Assessment profile
Assessment method- Setting
- research demonstration
- Exploitation
- demonstrated
- Evidence
- Grade C
- Remediation
- mitigation available
- AI role
- AGAINST AI
Basis and provenance
Scoped Claim(s) link to SRC-2026-DB233-GITHUB-MCP-PARTIAL-PR; reviewed as research in research_demonstration. Only reported controlled behavior or proof-of-concept is established; no production victim or impact magnitude is measured.
Assessed Nov 10, 2026 using diggingbeagle.assessment/1.
Evidence basis: CLM-DB233-GITHUB-MCP-PARTIAL-PR-01 · CLM-ENR234-DB233-GITHUB-MCP-PARTIAL-PR · github-mcp-server issue #3756
At a glance
No demonstrated malicious edits; this is a truthful-effect-receipt and retry failure.
Limits and uncertaintyFull account
Reviewer-assignment failure hid a successfully created pull request from the tool response.
Entry and authority. The agent can create a PR but requests an invalid or unassignable reviewer.
Observed mechanism and consequence. The repository issue shows PR creation succeeded before a 422 reviewer error made the tool return isError without the new PR ID, complicating safe retries.
Limits and status. No demonstrated malicious edits; this is a truthful-effect-receipt and retry failure.
The server could create a pull request and then fail to assign a reviewer, returning an error without the already committed PR identifier. A retry might report that a PR already exists even though the original caller never received its URL. The issue also describes partial edits during update operations. No attack or unauthorized reviewer assignment is needed for this failure: the tool's single success/failure bit misrepresented a multi-step transaction.
Mechanism and trust boundary
- 01
Precondition and trust context
The agent can create a PR but requests an invalid or unassignable reviewer.
- 02
Crossed boundary
The repository issue shows PR creation succeeded before a 422 reviewer error made the tool return isError without the new PR ID, complicating safe retries.
- 03
Observed limit
No demonstrated malicious edits; this is a truthful-effect-receipt and retry failure.
Timeline
- Nov 9, 2026Event type unspecified
Initial documented finding
GitHub MCP pull-request creation returned failure after committing a PR
- Nov 9, 2026disclosure
Public disclosure
Claims & evidence
CLM-DB233-GITHUB-MCP-PARTIAL-PR-01A repository issue reports: The repository issue shows PR creation succeeded before a 422 reviewer error made the tool return isError without the new PR ID, complicating safe retries.supported
Basis: reported finding
- supportsgithub-mcp-server issue #3756primary disclosure
Issue #3756, description and reproduction
CLM-ENR234-DB233-GITHUB-MCP-PARTIAL-PRA GitHub MCP pull request could be committed successfully before a subsequent reviewer-assignment failure caused the tool to return an error without the new PR ID.supported
Basis: reported finding
- supportsgithub-mcp-server issue #3756primary disclosure
Issue #3756, create_pull_request reproduction and partial side effects
Controls and mitigations
- Return structured partial-success receipts including committed PR IDs, changed fields and failed reviewer operations.
- Make retries idempotent by querying existing PRs before creation and distinguish a failed post-creation step from failure to create.
Unknowns and contradictions
- No demonstrated malicious edits; this is a truthful-effect-receipt and retry failure.
Sources and citation
Material revision history
- Nov 10, 2026 · Published version · first publication · revision 235