GHSA-RM8W-W798-G545
Vulnerability from github – Published: 2026-11-10 15:57 – Updated: 2026-11-10 15:57Impact
This is an incomplete fix for GHSA-jf3h-8jvx-vcrg. 0.32.0 made findings quote the ${NAME} a sibling .env filled in, instead of its value. The same release added a stop at 20,000 findings per document, and that stop is checked before the quoting step. A run that hits it reports the findings graded so far with the resolved values still in them, in text, JSON and SARIF output. That happens with or without --allow-partial-coverage.
The attack is the one GHSA-jf3h-8jvx-vcrg describes. In a CI workflow that writes secrets into the .env beside the Compose file, a pull request changes a line to image: "${DB_PASSWORD}". It also pushes the document past 20,000 findings, which a few hundred lines can do, because an aliased list is graded once for each service that uses it. The secret's value is quoted in a finding that reaches the job log, the JSON output and the SARIF uploaded to Code Scanning. GitHub masks registered secrets in the job log, but not in uploaded JSON or SARIF artifacts or in Code Scanning alert text. The run exits 2, or 0 with --allow-partial-coverage, after the output has been written.
The preconditions are the same as for GHSA-jf3h-8jvx-vcrg. The workflow writes secrets into the .env beside the Compose file and runs on pull requests from authors who should not see them, such as a pull_request_target workflow that lints the pull request's files. Pull requests from forks normally run without secrets, so the common setup is not exposed. The inflated document is visible in the pull request, but the workflow runs before anyone reviews it.
Found during an internal pre-1.0 security review.
Python package. compose-lint 0.32.0. Earlier releases have no finding limit, and their quoting gap is GHSA-jf3h-8jvx-vcrg itself.
GitHub Action. Action tag v0.32.0 is affected. Tags before v0.22.0 install the newest release unless version: is set, so they currently install 0.32.0.
pre-commit hook. rev: v0.32.0.
Docker image. composelint/compose-lint:0.32.0, :0.32 and latest are affected when the .env is mounted with the project.
Patches
No fixed release yet.
Workarounds
- Run with
--no-env, which skips the sibling.env. - Don't write secrets into the
.envbeside the Compose file in a job that lints untrusted pull requests.
Reproduction
project/.env DB_PASSWORD=example-secret (written by CI, not committed)
project/compose.yml x-p: &p [ ...200 port mappings... ]
services: one service with image: "${DB_PASSWORD}" and ports: *p,
plus about 100 more services with ports: *p
compose-lint --format sarif compose.yml exits 2 with "grading stopped at 20000 findings", and the reported findings contain example-secret. The same file with fewer services stays under the limit and quotes ${DB_PASSWORD}. Checked against the 0.32.0 wheel from PyPI, using a synthetic value.
{
"affected": [
{
"package": {
"ecosystem": "PyPI",
"name": "compose-lint"
},
"ranges": [
{
"events": [
{
"introduced": "0.32.0"
},
{
"fixed": "0.33.0"
}
],
"type": "ECOSYSTEM"
}
],
"versions": [
"0.32.0"
]
},
{
"package": {
"ecosystem": "GitHub Actions",
"name": "tmatens/compose-lint"
},
"ranges": [
{
"events": [
{
"introduced": "0.32.0"
},
{
"fixed": "0.33.0"
}
],
"type": "ECOSYSTEM"
}
],
"versions": [
"0.32.0"
]
}
],
"aliases": [
"CVE-2026-123796"
],
"database_specific": {
"cwe_ids": [
"CWE-200",
"CWE-532"
],
"github_reviewed": true,
"github_reviewed_at": "2026-11-10T15:57:22Z",
"nvd_published_at": "2026-11-09T18:17:26Z",
"severity": "LOW"
},
"details": "### Impact\n\nThis is an incomplete fix for GHSA-jf3h-8jvx-vcrg. 0.32.0 made findings quote the `${NAME}` a sibling `.env` filled in, instead of its value. The same release added a stop at 20,000 findings per document, and that stop is checked before the quoting step. A run that hits it reports the findings graded so far with the resolved values still in them, in text, JSON and SARIF output. That happens with or without `--allow-partial-coverage`.\n\nThe attack is the one GHSA-jf3h-8jvx-vcrg describes. In a CI workflow that writes secrets into the `.env` beside the Compose file, a pull request changes a line to `image: \"${DB_PASSWORD}\"`. It also pushes the document past 20,000 findings, which a few hundred lines can do, because an aliased list is graded once for each service that uses it. The secret\u0027s value is quoted in a finding that reaches the job log, the JSON output and the SARIF uploaded to Code Scanning. GitHub masks registered secrets in the job log, but not in uploaded JSON or SARIF artifacts or in Code Scanning alert text. The run exits 2, or 0 with `--allow-partial-coverage`, after the output has been written.\n\nThe preconditions are the same as for GHSA-jf3h-8jvx-vcrg. The workflow writes secrets into the `.env` beside the Compose file *and* runs on pull requests from authors who should not see them, such as a `pull_request_target` workflow that lints the pull request\u0027s files. Pull requests from forks normally run without secrets, so the common setup is not exposed. The inflated document is visible in the pull request, but the workflow runs before anyone reviews it.\n\nFound during an internal pre-1.0 security review.\n\n**Python package.** `compose-lint` 0.32.0. Earlier releases have no finding limit, and their quoting gap is GHSA-jf3h-8jvx-vcrg itself.\n\n**GitHub Action.** Action tag v0.32.0 is affected. Tags before v0.22.0 install the newest release unless `version:` is set, so they currently install 0.32.0.\n\n**pre-commit hook.** `rev: v0.32.0`.\n\n**Docker image.** `composelint/compose-lint:0.32.0`, `:0.32` and `latest` are affected when the `.env` is mounted with the project.\n\n### Patches\n\nNo fixed release yet.\n\n### Workarounds\n\n- Run with `--no-env`, which skips the sibling `.env`.\n- Don\u0027t write secrets into the `.env` beside the Compose file in a job that lints untrusted pull requests.\n\n### Reproduction\n\n```\nproject/.env DB_PASSWORD=example-secret (written by CI, not committed)\nproject/compose.yml x-p: \u0026p [ ...200 port mappings... ]\n services: one service with image: \"${DB_PASSWORD}\" and ports: *p,\n plus about 100 more services with ports: *p\n```\n\n`compose-lint --format sarif compose.yml` exits 2 with \"grading stopped at 20000 findings\", and the reported findings contain `example-secret`. The same file with fewer services stays under the limit and quotes `${DB_PASSWORD}`. Checked against the 0.32.0 wheel from PyPI, using a synthetic value.",
"id": "GHSA-rm8w-w798-g545",
"modified": "2026-11-10T15:57:22Z",
"published": "2026-11-10T15:57:22Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/tmatens/compose-lint/security/advisories/GHSA-rm8w-w798-g545"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-123796"
},
{
"type": "WEB",
"url": "https://github.com/tmatens/compose-lint/pull/978"
},
{
"type": "WEB",
"url": "https://github.com/tmatens/compose-lint/commit/31ec460399e06e98aa0d1cc578368a1960a8a0b3"
},
{
"type": "PACKAGE",
"url": "https://github.com/tmatens/compose-lint"
},
{
"type": "WEB",
"url": "https://github.com/tmatens/compose-lint/releases/tag/v0.33.0"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:4.0/AV:N/AC:L/AT:P/PR:N/UI:P/VC:L/VI:N/VA:N/SC:N/SI:N/SA:N",
"type": "CVSS_V4"
}
],
"summary": "compose-lint: Incomplete fix for GHSA-jf3h-8jvx-vcrg: a run stopped at the finding limit quotes a secret from a CI-generated `.env`"
}
Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
Vulnerability timeline
Sightings and recorded dates, grouped by day · UTC
How durations are calculated
Following the VTEM framework (Black Hat USA 2015), time to patch starts when the vendor or a coordinator was informed, total time to patch at discovery, and disclosure to patch at public disclosure; vendor response time runs from the report to its acknowledgement. Time to exploit and time unpatched start at the earliest discovery or report, falling back to public disclosure. Unpatched time ends at the first recorded patch, or the calculation time when no patch is recorded. Exploitation to patch shows the interval between those two events; a published proof of concept is a public exploit, not exploitation.
Each result shows its source dates. A negative interval means the end event came first; when date-only inputs leave the order open, the result says so. Date-only inputs are approximate. The earliest evidence wins: KEV report dates and sightings are labelled as such. Records describing another vulnerability are listed but never calculated from, and a PoC alone does not establish exploitation. Missing patch evidence does not prove that no patch exists. Filters below change the event list only.
Loading timeline…
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Related by attack behaviour
Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.