GHSA-65V8-Q222-64XX
Vulnerability from github – Published: 2026-11-10 15:57 – Updated: 2026-11-10 15:57Impact
compose-lint fix --apply converts the line numbers its fixers work with into positions in the file. The YAML parser treats a lone carriage return, U+0085, U+2028 and U+2029 as line breaks, but the conversion counted only \n. One of those characters inside a quoted value, which most editors and diff views don't show, shifted every later edit down by one line.
A Compose file from a pull request can use this to make a fix delete the line after the one it meant to change. For example, fixing CL-0014 (logging: driver: none) instead removes driver: none and the following networks: [internal], moving the service off an internal-only network onto the default bridge. The result is still valid Compose, every check fix runs after editing passes, and the command reports applied 1 fix(es) and exits 0. A later compose-lint check passes too, because no rule requires networks:. The dry-run diff (without --apply) does show the removed lines, but --apply prints only the note for the fix that was selected.
A related crash on the same files could empty the SARIF output for every file in the run. GitHub Code Scanning rejects an empty SARIF upload, which fails the step, so that part does not report a false clean result there.
Found during an internal pre-1.0 security review.
Docker image. composelint/compose-lint tags 0.9.0 through 0.17.0, and the matching minor tags, are affected when used to run fix --apply. Fixed from 0.18.0.
The GitHub Action does not run fix and is not affected.
Patches
Fixed in 0.18.0. Line splitting and position conversion now share one definition of a line break, so they can't disagree. A Compose file containing an ambiguous line break (a lone \r, U+0085, U+2028 or U+2029) is refused with exit code 2, because no line numbering is correct for both the parser and the editor. CRLF and LF files are unaffected.
Workarounds
- Upgrade to 0.18.0 or later.
- On older versions, review the dry-run diff before
--apply, and don't runfix --applyover files from untrusted pull requests.
Reproduction
services:
web:
image: nginx@sha256:<digest>
labels:
note: "line one<U+2028>line two"
logging:
driver: none
networks: [internal]
networks:
internal:
internal: true
compose-lint fix --apply compose.yml on 0.17.0 reports applied 1 fix(es), exits 0, and removes both driver: none and networks: [internal].
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 0.17.0"
},
"package": {
"ecosystem": "PyPI",
"name": "compose-lint"
},
"ranges": [
{
"events": [
{
"introduced": "0.9.0"
},
{
"fixed": "0.18.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-123370"
],
"database_specific": {
"cwe_ids": [
"CWE-193"
],
"github_reviewed": true,
"github_reviewed_at": "2026-11-10T15:57:02Z",
"nvd_published_at": "2026-11-09T18:17:25Z",
"severity": "MODERATE"
},
"details": "### Impact\n\n`compose-lint fix --apply` converts the line numbers its fixers work with into positions in the file. The YAML parser treats a lone carriage return, U+0085, U+2028 and U+2029 as line breaks, but the conversion counted only `\\n`. One of those characters inside a quoted value, which most editors and diff views don\u0027t show, shifted every later edit down by one line.\n\nA Compose file from a pull request can use this to make a fix delete the line after the one it meant to change. For example, fixing CL-0014 (`logging: driver: none`) instead removes `driver: none` and the following `networks: [internal]`, moving the service off an internal-only network onto the default bridge. The result is still valid Compose, every check `fix` runs after editing passes, and the command reports `applied 1 fix(es)` and exits 0. A later `compose-lint check` passes too, because no rule requires `networks:`. The dry-run diff (without `--apply`) does show the removed lines, but `--apply` prints only the note for the fix that was selected.\n\nA related crash on the same files could empty the SARIF output for every file in the run. GitHub Code Scanning rejects an empty SARIF upload, which fails the step, so that part does not report a false clean result there.\n\nFound during an internal pre-1.0 security review.\n\n**Docker image.** `composelint/compose-lint` tags 0.9.0 through 0.17.0, and the matching minor tags, are affected when used to run `fix --apply`. Fixed from `0.18.0`.\n\nThe GitHub Action does not run `fix` and is not affected.\n\n### Patches\n\nFixed in 0.18.0. Line splitting and position conversion now share one definition of a line break, so they can\u0027t disagree. A Compose file containing an ambiguous line break (a lone `\\r`, U+0085, U+2028 or U+2029) is refused with exit code 2, because no line numbering is correct for both the parser and the editor. CRLF and LF files are unaffected.\n\n### Workarounds\n\n- Upgrade to 0.18.0 or later.\n- On older versions, review the dry-run diff before `--apply`, and don\u0027t run `fix --apply` over files from untrusted pull requests.\n\n### Reproduction\n\n```yaml\nservices:\n web:\n image: nginx@sha256:\u003cdigest\u003e\n labels:\n note: \"line one\u003cU+2028\u003eline two\"\n logging:\n driver: none\n networks: [internal]\nnetworks:\n internal:\n internal: true\n```\n\n`compose-lint fix --apply compose.yml` on 0.17.0 reports `applied 1 fix(es)`, exits 0, and removes both `driver: none` and `networks: [internal]`.",
"id": "GHSA-65v8-q222-64xx",
"modified": "2026-11-10T15:57:02Z",
"published": "2026-11-10T15:57:02Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/tmatens/compose-lint/security/advisories/GHSA-65v8-q222-64xx"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-123370"
},
{
"type": "WEB",
"url": "https://github.com/tmatens/compose-lint/pull/547"
},
{
"type": "WEB",
"url": "https://github.com/tmatens/compose-lint/pull/549"
},
{
"type": "WEB",
"url": "https://github.com/tmatens/compose-lint/commit/c9d453a6a69c076788fe1db05cb54684a616d1de"
},
{
"type": "WEB",
"url": "https://github.com/tmatens/compose-lint/commit/cbbf886d91347dea9c51b584eb50088e70007d0d"
},
{
"type": "PACKAGE",
"url": "https://github.com/tmatens/compose-lint"
},
{
"type": "WEB",
"url": "https://github.com/tmatens/compose-lint/releases/tag/v0.18.0"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:A/VC:N/VI:L/VA:N/SC:N/SI:L/SA:N",
"type": "CVSS_V4"
}
],
"summary": "compose-lint: `fix --apply` can delete a line the user did not select, including a security setting"
}
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.