GHSA-GJ7R-P794-HHQ5
Vulnerability from github – Published: 2026-11-16 22:18 – Updated: 2026-11-16 22:18Ecosystem / package: Go — github.com/kubeshark/kubeshark (the kubeshark
CLI, kubeshark mcp subcommand)
Affected component: cmd/mcpRunner.go, MCP tool download_file
Affected versions: >= v53.1.0, < v53.4.0 (the flaw shipped with the
tool's first release; v53.3.0 was the last affected release)
Patched version: v53.4.0
Credit: novice-22 (Urim), via private coordinated disclosure.
Summary
The download_file MCP tool fetches a caller-supplied path from the Kubeshark
Hub API and writes the response body to a caller-supplied dest on the local
filesystem with no validation of dest. An attacker who can influence the MCP
tool arguments — e.g. via indirect prompt injection against an AI agent driving
kubeshark mcp — can write attacker-controlled bytes to an arbitrary path
writable by the user running the MCP server (e.g. ~/.ssh/authorized_keys,
~/.kube/config, ~/.bashrc), and can request arbitrary Hub API endpoints.
download_file is registered unconditionally (available in the default,
non---allow-destructive posture).
Introduced in
Commit eb7dc42b — "Add get_file_url and download_file MCP tools (#1853)",
2026-03-04 — first released in v53.1.0 (v53.1.0~1). An earlier attempt
(dbfd17d9) was reverted (a46f05c4) before any release, so no tag before
v53.1.0 is affected.
Details
In callDownloadFile (cmd/mcpRunner.go), both path and dest come straight
from the tool arguments:
- dest was passed directly to os.Create(dest) with no confinement — no
filepath.Clean, no base-directory containment, no rejection of absolute
paths or ../ traversal. Any path the process can write is reachable.
- path was concatenated onto the Hub API base URL with no allowlist, so any
Hub endpoint (including captured-traffic exports) is reachable.
Auth context (CWE-306 / CWE-200)
The tool originally issued the fetch with no authentication. The authentication
gap was addressed by the gated-Hub authentication refactor: the download now
forwards the operator's Hub credentials (ServiceAccount token / License-Key) and
rejects auth-required responses instead of writing a login page to disk. But
the Hub ships auth-disabled by default (community/dev), so on a default
kubeshark tap the Hub API answers without authentication and captured traffic
(credentials/tokens/PII) can still be read and exfiltrated end-to-end. Enforcing
Hub auth by default is tracked separately (gated-auth workstream) and is not
part of this advisory's fix.
Proof of concept (reporter, abridged)
Following the tool's documented workflow on a default community install:
export_snapshot_pcap → obtain a relative path → download_file with that
path and an attacker-chosen dest. The reporter retrieved a captured-traffic
PCAP and recovered a planted canary Authorization: Bearer … credential from
it, and demonstrated writing the fetched bytes to an arbitrary local path.
Impact
- Integrity (local): arbitrary file write on the machine running
kubeshark mcp, as that user. - Confidentiality: read of Hub API endpoints, incl. captured network traffic (credentials, tokens, PII) on a default (auth-disabled) install.
- Reliable RCE is not demonstrated (script endpoints are JSON-wrapped).
Severity rationale
Kubeshark scores this 5.3 Medium, down from the reporter's 7.1, on two metrics:
AV:N→AV:L. The vulnerable component has no network listener the attacker reaches.kubeshark mcpis a local stdio MCP server; the attacker's input arrives as content an agent running on that host ingests and then replays into a local tool call. The delivery of the poisoned content may cross a network, but the attack surface being exploited is local to the machine.I:H→I:L. The write is confined to paths writable by the user runningkubeshark mcp(normally unprivileged) — no system-file or other-user modification. The attacker also controls where the bytes land, not what they are: the body comes from a Hub API response, so the content is constrained to what the Hub will emit. Reliable RCE was not demonstrated.A:L→A:N. Nothing in the issue degrades availability of the CLI, the Hub, or the host; the originalA:Lis unsupported.
Kubeshark kept C:H: on a default community install the Hub API answers
unauthenticated, so captured traffic (credentials, tokens, PII) is genuinely
readable end-to-end. Kubeshark kept AC:H (the attacker must land an injection in
content the agent ingests and know the deployment layout), PR:N
(download_file is registered unconditionally — no --allow-destructive
needed), UI:R (an operator must be driving an agent against
kubeshark mcp), and S:U (the impact stays within the local user's
authorization scope).
Patches
Fixed in v53.4.0 by commit
a76a1920
(PR #1957).
Changes in cmd/mcpRunner.go:
- dest is resolved through secureDownloadDest and confined to a base
directory (the CWD by default, relocatable via KUBESHARK_MCP_DOWNLOAD_DIR);
a dest that resolves outside the base (via ../ or an absolute path) is
rejected.
- path is rejected if it contains .. segments.
- Containment is re-checked after resolving symlinks, so a symlinked
subdirectory inside the base (or a dest that is itself a symlink) cannot
redirect the write outside it; the file is opened with O_NOFOLLOW where the
platform provides it.
- Credential forwarding + auth-required detection on the fetch (shipped in the
gated-auth refactor).
Workarounds
- Run
kubeshark mcpas an unprivileged user in an isolated working directory. - Do not expose the MCP server to untrusted content / untrusted agent inputs.
- Enable Hub authentication (
AUTH_ENABLED=true+ an IdP/license) to close the unauthenticated read path.
Follow-ups (not blocking this advisory)
- Strict prefix allowlist for the Hub
path(current fix only rejects..). - Hub-side default auth enforcement (gated-auth workstream).
{
"affected": [
{
"package": {
"ecosystem": "Go",
"name": "github.com/kubeshark/kubeshark"
},
"ranges": [
{
"events": [
{
"introduced": "0.0.2-0.20260304171723-eb7dc42b6ea0"
},
{
"fixed": "0.0.2-0.20260812200004-a76a1920dc75"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-129235"
],
"database_specific": {
"cwe_ids": [
"CWE-22",
"CWE-88"
],
"github_reviewed": true,
"github_reviewed_at": "2026-11-16T22:18:58Z",
"nvd_published_at": "2026-11-16T17:17:31Z",
"severity": "MODERATE"
},
"details": "**Ecosystem / package:** Go \u2014 `github.com/kubeshark/kubeshark` (the `kubeshark`\nCLI, `kubeshark mcp` subcommand)\n**Affected component:** `cmd/mcpRunner.go`, MCP tool `download_file`\n**Affected versions:** `\u003e= v53.1.0`, `\u003c v53.4.0` (the flaw shipped with the\ntool\u0027s first release; v53.3.0 was the last affected release)\n**Patched version:** `v53.4.0`\n**Credit:** novice-22 (Urim), via private coordinated disclosure.\n\n## Summary\nThe `download_file` MCP tool fetches a caller-supplied `path` from the Kubeshark\nHub API and writes the response body to a caller-supplied `dest` on the local\nfilesystem with no validation of `dest`. An attacker who can influence the MCP\ntool arguments \u2014 e.g. via indirect prompt injection against an AI agent driving\n`kubeshark mcp` \u2014 can write attacker-controlled bytes to an arbitrary path\nwritable by the user running the MCP server (e.g. `~/.ssh/authorized_keys`,\n`~/.kube/config`, `~/.bashrc`), and can request arbitrary Hub API endpoints.\n`download_file` is registered unconditionally (available in the default,\nnon-`--allow-destructive` posture).\n\n## Introduced in\nCommit `eb7dc42b` \u2014 \"Add get_file_url and download_file MCP tools (#1853)\",\n2026-03-04 \u2014 first released in **v53.1.0** (`v53.1.0~1`). An earlier attempt\n(`dbfd17d9`) was reverted (`a46f05c4`) before any release, so no tag before\nv53.1.0 is affected.\n\n## Details\nIn `callDownloadFile` (`cmd/mcpRunner.go`), both `path` and `dest` come straight\nfrom the tool arguments:\n- `dest` was passed directly to `os.Create(dest)` with no confinement \u2014 no\n `filepath.Clean`, no base-directory containment, no rejection of absolute\n paths or `../` traversal. Any path the process can write is reachable.\n- `path` was concatenated onto the Hub API base URL with no allowlist, so any\n Hub endpoint (including captured-traffic exports) is reachable.\n\n## Auth context (CWE-306 / CWE-200)\nThe tool originally issued the fetch with no authentication. The authentication\ngap was addressed by the gated-Hub authentication refactor: the download now\nforwards the operator\u0027s Hub credentials (ServiceAccount token / License-Key) and\nrejects auth-required responses instead of writing a login page to disk. **But**\nthe Hub ships auth-disabled by default (community/dev), so on a default\n`kubeshark tap` the Hub API answers without authentication and captured traffic\n(credentials/tokens/PII) can still be read and exfiltrated end-to-end. Enforcing\nHub auth by default is tracked separately (gated-auth workstream) and is **not**\npart of this advisory\u0027s fix.\n\n## Proof of concept (reporter, abridged)\nFollowing the tool\u0027s documented workflow on a default community install:\n`export_snapshot_pcap` \u2192 obtain a relative path \u2192 `download_file` with that\n`path` and an attacker-chosen `dest`. The reporter retrieved a captured-traffic\nPCAP and recovered a planted canary `Authorization: Bearer \u2026` credential from\nit, and demonstrated writing the fetched bytes to an arbitrary local path.\n\n## Impact\n- **Integrity (local):** arbitrary file write on the machine running\n `kubeshark mcp`, as that user.\n- **Confidentiality:** read of Hub API endpoints, incl. captured network traffic\n (credentials, tokens, PII) on a default (auth-disabled) install.\n- Reliable RCE is not demonstrated (script endpoints are JSON-wrapped).\n\n## Severity rationale\nKubeshark scores this **5.3 Medium**, down from the reporter\u0027s 7.1, on two metrics:\n\n- **`AV:N` \u2192 `AV:L`.** The vulnerable component has no network listener the\n attacker reaches. `kubeshark mcp` is a local stdio MCP server; the attacker\u0027s\n input arrives as content an agent running on that host ingests and then\n replays into a local tool call. The delivery of the poisoned content may cross\n a network, but the attack surface being exploited is local to the machine.\n- **`I:H` \u2192 `I:L`.** The write is confined to paths writable by the user running\n `kubeshark mcp` (normally unprivileged) \u2014 no system-file or other-user\n modification. The attacker also controls *where* the bytes land, not *what*\n they are: the body comes from a Hub API response, so the content is\n constrained to what the Hub will emit. Reliable RCE was not demonstrated.\n- **`A:L` \u2192 `A:N`.** Nothing in the issue degrades availability of the CLI, the\n Hub, or the host; the original `A:L` is unsupported.\n\nKubeshark kept **`C:H`**: on a default community install the Hub API answers\nunauthenticated, so captured traffic (credentials, tokens, PII) is genuinely\nreadable end-to-end. Kubeshark kept **`AC:H`** (the attacker must land an injection in\ncontent the agent ingests and know the deployment layout), **`PR:N`**\n(`download_file` is registered unconditionally \u2014 no `--allow-destructive`\nneeded), **`UI:R`** (an operator must be driving an agent against\n`kubeshark mcp`), and **`S:U`** (the impact stays within the local user\u0027s\nauthorization scope).\n\n## Patches\nFixed in **v53.4.0** by commit\n[`a76a1920`](https://github.com/kubeshark/kubeshark/commit/a76a1920dc7587be700a17f1592f58a9512ad214)\n(PR [#1957](https://github.com/kubeshark/kubeshark/pull/1957)).\n\nChanges in `cmd/mcpRunner.go`:\n- `dest` is resolved through `secureDownloadDest` and confined to a base\n directory (the CWD by default, relocatable via `KUBESHARK_MCP_DOWNLOAD_DIR`);\n a `dest` that resolves outside the base (via `../` or an absolute path) is\n rejected.\n- `path` is rejected if it contains `..` segments.\n- Containment is re-checked after resolving symlinks, so a symlinked\n subdirectory inside the base (or a `dest` that is itself a symlink) cannot\n redirect the write outside it; the file is opened with `O_NOFOLLOW` where the\n platform provides it.\n- Credential forwarding + auth-required detection on the fetch (shipped in the\n gated-auth refactor).\n\n## Workarounds\n- Run `kubeshark mcp` as an unprivileged user in an isolated working directory.\n- Do not expose the MCP server to untrusted content / untrusted agent inputs.\n- Enable Hub authentication (`AUTH_ENABLED=true` + an IdP/license) to close the\n unauthenticated read path.\n\n## Follow-ups (not blocking this advisory)\n- Strict prefix allowlist for the Hub `path` (current fix only rejects `..`).\n- Hub-side default auth enforcement (gated-auth workstream).",
"id": "GHSA-gj7r-p794-hhq5",
"modified": "2026-11-16T22:18:58Z",
"published": "2026-11-16T22:18:58Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/kubeshark/kubeshark/security/advisories/GHSA-gj7r-p794-hhq5"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-129235"
},
{
"type": "WEB",
"url": "https://github.com/kubeshark/kubeshark/pull/1957"
},
{
"type": "WEB",
"url": "https://github.com/kubeshark/kubeshark/commit/a76a1920dc7587be700a17f1592f58a9512ad214"
},
{
"type": "PACKAGE",
"url": "https://github.com/kubeshark/kubeshark"
},
{
"type": "WEB",
"url": "https://github.com/kubeshark/kubeshark/releases/tag/v53.4.0"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:H/PR:N/UI:R/S:U/C:H/I:L/A:N",
"type": "CVSS_V3"
}
],
"summary": "Kubeshark: MCP `download_file` writes fetched bytes to an unvalidated destination path"
}
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.
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.