pull_request_target

How pull_request_target works

pull_request_target is unusual because it evaluates the workflow from the base repository, not the contributor’s fork. That means the workflow can access repository secrets and write-scoped permissions even when the code being proposed comes from an untrusted source.

Its security significance comes from that trust boundary. A workflow that looks harmless on the surface can still become a privileged execution path if it checks out, runs, or interprets attacker-controlled content too early.

Why it is dangerous in GitHub Actions

The main danger is that the trigger separates where the workflow runs from who supplied the code. If a job uses pull request contents before it has established trust, the attacker may influence scripts, metadata, caches, or generated files while the job still has access to protected resources.

This is why labels such as “safe” or “approved” must be treated as authorization signals, not as proof that the code itself is trustworthy. In practice, the risky pattern is not the trigger alone, but the combination of privileged context, untrusted input, and workflow steps that read or execute that input.

To understand the broader pattern of CI/CD abuse and why repository workflows need tighter control than ordinary app code, see the Top 10 CI/CD Security Risks.

How teams check exposure and contain blast radius

Exposure starts with inventory: identify every workflow that uses pull_request_target, then trace whether it can reach secrets, deployment credentials, signing keys, package publishing tokens, or self-hosted runners. The next question is whether the workflow processes forked content, generated diffs, labels, comments, artifacts, or branch data before any trust decision is made.

Teams should also review repository protections that shape how the trigger is used, especially branch protections, required reviews, code ownership checks, and permissions on the default GITHUB_TOKEN. A safe design keeps the privileged workflow path as small as possible and avoids giving the job more authority than the validation task actually needs.

GitHub-specific hardening guidance belongs in the GitHub and GitHub Actions Security Guide, which covers token scope, action pinning, runner hardening, and workflow injection.

Detection, review, and recovery after misuse

If a pull_request_target workflow may have been abused, the review should focus on whether secrets, tokens, or write-capable credentials were exposed or used unexpectedly. Look for suspicious workflow edits, unexpected job steps, repository writes, secret access, artifact tampering, and downstream package or release changes that do not match normal contributor behavior.

Post-incident validation should also include replaying the workflow logic with a clean understanding of what input came from the fork and what input came from the base repository. In many cases, the compromise is not obvious from a single run, because the attacker may use the workflow to plant a later supply-chain action rather than trigger an immediate obvious failure.

When investigation turns into credential hygiene, the most relevant recovery lens is secret and token exposure, including revocation and rotation of anything the workflow could reach. The JetBrains GitHub plugin token exposure and Travis CI secrets exposure 2021 pages are useful examples of how exposed tokens should be treated as compromised until proven otherwise.

Risk and Threat Considerations

pull_request_target creates a high-value trust boundary for attackers because it can turn a contributor-controlled pull request into privileged repository execution. The practical risk is secret exposure, unauthorized writes, and supply-chain abuse when workflow logic trusts unreviewed input too early.

Failure mechanism: An attacker shapes workflow inputs, labels, paths, or generated content so a privileged job executes malicious instructions or discloses secrets before the maintainer has validated the request.

Impact: The result can be credential theft, poisoned releases, repository tampering, or lateral movement into downstream CI/CD and package publishing systems.

Practitioner Guidance

Governance implication: Treat pull_request_target as a privileged trigger that needs an explicit approval model, not as a default pull request setting. Limit it to workflows that must operate in base-repository context, and keep those workflows narrow enough that they can validate safely without exposing sensitive material.

Practitioner takeaway: If a workflow can be built to validate untrusted pull requests without secret access, prefer that design; reserve pull_request_target for cases where the privileged context is truly required.

Related resources from NHI Mgmt Group

• Should organisations allow pull_request_target for automated dependency workflows?
• What do security teams get wrong about pull_request_target workflows?
• Why do pull_request_target workflows create more risk than standard pull request workflows?
• What breaks when pull_request_target is used to run untrusted code in GitHub Actions?