CVE-2021-32638 | CodeQL runner: Command-line options that make GitHub access tokens visible to other processes are now deprecated

Exp

Github's CodeQL action is provided to run CodeQL-based code scanning on non-GitHub CI/CD systems and requires a GitHub access token to connect to a GitHub repository. The runner and its documentation previously suggested passing the GitHub token as a command-line parameter to the process instead of reading it from a file, standard input, or an environment variable. This approach made the token visible to other processes on the same machine, for example in the output of the `ps` command. If the CI system publicly exposes the output of `ps`, for example by logging the output, then the GitHub access token can be exposed beyond the scope intended. Users of the CodeQL runner on 3rd-party systems, who are passing a GitHub token via the `--github-auth` flag, are affected. This applies to both GitHub.com and GitHub Enterprise users. Users of the CodeQL Action on GitHub Actions are not affected. The `--github-auth` flag is now considered insecure and deprecated. The undocumented `--external-repository-token` flag has been removed. To securely provide a GitHub access token to the CodeQL runner, users should **do one of the following instead**: Use the `--github-auth-stdin` flag and pass the token on the command line via standard input OR set the `GITHUB_TOKEN` environment variable to contain the token, then call the command without passing in the token. The old flag remains present for backwards compatibility with existing workflows. If the user tries to specify an access token using the `--github-auth` flag, there is a deprecation warning printed to the terminal that directs the user to one of the above options. All CodeQL runner releases codeql-bundle-20210304 onwards contain the patches. We recommend updating to a recent version of the CodeQL runner, storing a token in your CI system's secret storage mechanism, and passing the token to the CodeQL runner using `--github-auth-stdin` or the `GITHUB_TOKEN` environment variable. If still using the old flag, ensure that process output, such as from `ps`, is not persisted in CI logs.

Published: 2021-05-25 Last update: 2024-11-21 Assigner: [email protected] Source: [email protected]

Conclusion & alert: CVE-2021-32638 is rated Exploit Available (50/100): CVSS Medium severity, with low exploitation likelihood (EPSS 0.06%). Core evidence: 1 public exploit reference(s) are indexed (Exploit-DB). Mandatory action: Public exploits are available—assess exposure, apply mitigations, and prioritize patching.

Risk is dynamic; we continuously reassess and refresh what is shown on this page as upstream context changes.

Public exploit references (Exploit-DB) for CVE-2021-32638

EDB-ID Source Kind Published Link
nvd_ref exploit_tag Exploit-DB ↗

Exploit prediction scoring system (EPSS) score for CVE-2021-32638

EPSS lead: Daily EPSS estimates relative likelihood of exploitation; percentile ranks this CVE among scored vulnerabilities (higher = more severe relative rank).

# Date Old EPSS score New EPSS score Delta (New - Old)
1 2023-03-07 0.95% 0.06% -0.89%
2 2022-04-01 1.67% 0.95% -0.72%
3 2022-02-04 1.67%

Full EPSS history (7 records total)

Common vulnerability scoring system (CVSS) metrics for CVE-2021-32638

CVSS metrics for this CVE.

Base score Version Severity Vector Exploitability Impact Score source
4.4 3.1 MEDIUM
CVSS:3.1/AV:L/AC:L/PR:H/UI:N/S:U/C:H/I:N/A:N Click to expand
Attack vector (AV:L)
They already need access on the box, or another person has to do something wrong; it’s not a remote drive-by.
Attack complexity (AC:L)
Once they can reach the bug, pulling it off is straightforward—no weird race conditions or rare setup.
Privileges required (PR:H)
They need powerful rights—admin, root, or similar—before this pays off.
User interaction (UI:N)
Nobody has to click “OK” or open a trap file; it can work without a victim helping.
Scope (S:U)
Damage stays in the same “trust bubble” as the broken component—no big spill into unrelated systems.
Confidentiality (C:H)
Serious risk that confidential data gets exposed in a big way.
Integrity (I:N)
Data isn’t meaningfully altered or forged.
Availability (A:N)
Service keeps running; no real outage angle.
0.8 3.6 [email protected]
4.4 3.1 MEDIUM
CVSS:3.1/AV:L/AC:L/PR:H/UI:N/S:U/C:H/I:N/A:N Click to expand
Attack vector (AV:L)
They already need access on the box, or another person has to do something wrong; it’s not a remote drive-by.
Attack complexity (AC:L)
Once they can reach the bug, pulling it off is straightforward—no weird race conditions or rare setup.
Privileges required (PR:H)
They need powerful rights—admin, root, or similar—before this pays off.
User interaction (UI:N)
Nobody has to click “OK” or open a trap file; it can work without a victim helping.
Scope (S:U)
Damage stays in the same “trust bubble” as the broken component—no big spill into unrelated systems.
Confidentiality (C:H)
Serious risk that confidential data gets exposed in a big way.
Integrity (I:N)
Data isn’t meaningfully altered or forged.
Availability (A:N)
Service keeps running; no real outage angle.
0.8 3.6 [email protected]
2.1 2.0 LOW
AV:L/AC:L/Au:N/C:P/I:N/A:N Click to expand
Access vector (AV:L)
Requires local access to the target system.
Access complexity (AC:L)
Exploitation conditions are straightforward and predictable.
Authentication (AU:N)
No authentication is required.
Confidentiality impact (C:P)
Partial confidentiality impact.
Integrity impact (I:N)
No integrity impact.
Availability impact (A:N)
No availability impact.
3.9 2.9 [email protected]

Weakness enumeration for CVE-2021-32638

Affected software / configurations for CVE-2021-32638

Vendor Product Version Raw CPE
github codeql_action < 20210304 cpe:2.3:a:github:codeql_action:*:*:*:*:*:*:*:*

References for CVE-2021-32638

cvelogic Threat Intelligence