GHSA-756X-9HF6-Q4H4
Vulnerability from github – Published: 2026-09-02 21:54 – Updated: 2026-09-02 21:54Summary
An authenticated user can upload a crafted agent bundle that defines a server-side Python callable tool. The server validates the uploaded bundle, but it does not block dangerous callable: paths in untrusted user-provided agent configs.
When the tool is invoked, the runner imports and executes that Python callable. A crafted bundle can point the tool at subprocess.check_output, which lets the attacker run a local command on the runner machine.
This is serious because hosted Omnigent deployments can have multiple users and shared or managed runner hosts. A normal logged-in user should not be able to make the runner execute arbitrary local commands.
Details
The uploaded bundle path is reachable through session creation. Omnigent supports custom agent bundles through multipart POST /v1/sessions; the Web UI uses this same endpoint for custom agents:
Create custom agent builds and uploads a bundle
\
createBundledSession posts the bundle to /v1/sessions
A crafted bundle can also be submitted directly through the API/CLI when creating a session from agent YAML. The PoC below uses this direct bundle shape so the vulnerable trust boundary is easy to reproduce.
On the server, uploaded bundles are validated here:
validate_agent_bundle
The validator already treats uploaded bundles as untrusted in some ways. For example, it disables server-side environment expansion and can enforce a policy-handler allowlist:
spec = load(
bundle_bytes,
dest=Path(tmpdir) / "agent",
expand_env=False,
enforce_handler_allowlist=enforce_handler_allowlist,
)
However, the same validation does not reject tools.<name>.callable entries in an uploaded agent bundle. Python callable tools are an intended trusted/operator feature, but the upload path accepts them from tenant-provided bundles too.
The docs describe server-side Python callable tools here: Python function tool docs
\
Later, the runner resolves and executes those callables: _resolve_spec_callable imports the dotted path
mod = importlib.import_module(module_name)
fn = getattr(mod, attr_name, None)
_execute_spec_callable_tool calls the resolved function
result = await asyncio.to_thread(resolved, **args)
The high-level tool dispatcher falls back to this callable execution path: execute_tool callable fallback
output = await _execute_spec_callable_tool(tool_name, args, agent_spec=agent_spec)
The problem is the trust boundary. Python callables are a trusted local developer feature, but uploaded agent bundles can be user-controlled in a hosted/multi-user server. The upload validator blocks some trusted-only policy behavior, but it does not block trusted-only tool callables.
Suggested fix: when validating HTTP-uploaded bundles on a shared/multi-user server, reject server-side Python callable tools unless they are explicitly allowlisted by the operator. A conservative fix would reject callable: tool paths for tenant-uploaded bundles, while keeping trusted local/operator-authored configs working.
Impact
This is authenticated RCE against the Omnigent runner environment. A normal authenticated user who can upload or create a custom agent bundle can declare a server-side Python callable and have the runner import and execute it.
Because the callable runs inside the runner process, the attacker-controlled code runs with the runner's permissions. In a hosted or multi-user deployment, this can expose runner-accessible files, environment variables, credentials, workspace data, internal services reachable from the runner, and runner availability.
This is high impact because it crosses the boundary from "upload an agent config" to "execute local code on shared runner infrastructure" without requiring admin/operator access.
{
"affected": [
{
"package": {
"ecosystem": "PyPI",
"name": "omnigent"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "0.3.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-62675"
],
"database_specific": {
"cwe_ids": [
"CWE-94"
],
"github_reviewed": true,
"github_reviewed_at": "2026-09-02T21:54:52Z",
"nvd_published_at": "2026-08-21T18:16:49Z",
"severity": "HIGH"
},
"details": "### Summary\n\nAn authenticated user can upload a crafted agent bundle that defines a server-side Python callable tool. The server validates the uploaded bundle, but it does not block dangerous `callable:` paths in untrusted user-provided agent configs.\n\nWhen the tool is invoked, the runner imports and executes that Python callable. A crafted bundle can point the tool at `subprocess.check_output`, which lets the attacker run a local command on the runner machine.\n\nThis is serious because hosted Omnigent deployments can have multiple users and shared or managed runner hosts. A normal logged-in user should not be able to make the runner execute arbitrary local commands.\n\n### Details\n\nThe uploaded bundle path is reachable through session creation. Omnigent supports custom agent bundles through multipart `POST /v1/sessions`; the Web UI uses this same endpoint for custom agents:\n\n[`Create custom agent` builds and uploads a bundle](https://github.com/omnigent-ai/omnigent/blob/23d42d9a7d46a3b6a0b40d5bf4125b91193bf831/ap-web/src/shell/NewChatDialog.tsx#L1265-L1277) \n\\\n[`createBundledSession` posts the bundle to `/v1/sessions`](https://github.com/omnigent-ai/omnigent/blob/23d42d9a7d46a3b6a0b40d5bf4125b91193bf831/ap-web/src/lib/sessionsApi.ts#L421-L460)\n\nA crafted bundle can also be submitted directly through the API/CLI when creating a session from agent YAML. The PoC below uses this direct bundle shape so the vulnerable trust boundary is easy to reproduce.\n\nOn the server, uploaded bundles are validated here:\n[`validate_agent_bundle`](https://github.com/omnigent-ai/omnigent/blob/23d42d9a7d46a3b6a0b40d5bf4125b91193bf831/omnigent/server/bundles.py#L13-L75)\n\nThe validator already treats uploaded bundles as untrusted in some ways. For example, it disables server-side environment expansion and can enforce a policy-handler allowlist:\n```python\nspec = load(\n bundle_bytes,\n dest=Path(tmpdir) / \"agent\",\n expand_env=False,\n enforce_handler_allowlist=enforce_handler_allowlist,\n)\n```\nHowever, the same validation does not reject `tools.\u003cname\u003e.callable` entries in an uploaded agent bundle. Python callable tools are an intended trusted/operator feature, but the upload path accepts them from tenant-provided bundles too.\n\nThe docs describe server-side Python callable tools here: [`Python function tool` docs](https://github.com/omnigent-ai/omnigent/blob/23d42d9a7d46a3b6a0b40d5bf4125b91193bf831/docs/AGENT_YAML_SPEC.md#L229-L246)\n\\\nLater, the runner resolves and executes those callables: [`_resolve_spec_callable` imports the dotted path](https://github.com/omnigent-ai/omnigent/blob/23d42d9a7d46a3b6a0b40d5bf4125b91193bf831/omnigent/runner/tool_dispatch.py#L444-L481)\n```python\nmod = importlib.import_module(module_name)\nfn = getattr(mod, attr_name, None)\n```\n[`_execute_spec_callable_tool` calls the resolved function](https://github.com/omnigent-ai/omnigent/blob/23d42d9a7d46a3b6a0b40d5bf4125b91193bf831/omnigent/runner/tool_dispatch.py#L484-L511)\n```python\nresult = await asyncio.to_thread(resolved, **args)\n```\nThe high-level tool dispatcher falls back to this callable execution path: [`execute_tool` callable fallback](https://github.com/omnigent-ai/omnigent/blob/23d42d9a7d46a3b6a0b40d5bf4125b91193bf831/omnigent/runner/tool_dispatch.py#L3958-L3965)\n```python\noutput = await _execute_spec_callable_tool(tool_name, args, agent_spec=agent_spec)\n```\n\nThe problem is the trust boundary. Python callables are a trusted local developer feature, but uploaded agent bundles can be user-controlled in a hosted/multi-user server. The upload validator blocks some trusted-only policy behavior, but it does not block trusted-only tool callables.\n\nSuggested fix: when validating HTTP-uploaded bundles on a shared/multi-user server, reject server-side Python callable tools unless they are explicitly allowlisted by the operator. A conservative fix would reject `callable:` tool paths for tenant-uploaded bundles, while keeping trusted local/operator-authored configs working.\n\n### Impact\n\nThis is authenticated RCE against the Omnigent runner environment. A normal authenticated user who can upload or create a custom agent bundle can declare a server-side Python callable and have the runner import and execute it.\n\nBecause the callable runs inside the runner process, the attacker-controlled code runs with the runner\u0027s permissions. In a hosted or multi-user deployment, this can expose runner-accessible files, environment variables, credentials, workspace data, internal services reachable from the runner, and runner availability.\n\nThis is high impact because it crosses the boundary from \"upload an agent config\" to \"execute local code on shared runner infrastructure\" without requiring admin/operator access.",
"id": "GHSA-756x-9hf6-q4h4",
"modified": "2026-09-02T21:54:52Z",
"published": "2026-09-02T21:54:52Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/omnigent-ai/omnigent/security/advisories/GHSA-756x-9hf6-q4h4"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-62675"
},
{
"type": "WEB",
"url": "https://github.com/omnigent-ai/omnigent/pull/1430"
},
{
"type": "WEB",
"url": "https://github.com/omnigent-ai/omnigent/commit/1f3f398f41cbf97b905133c21e848621c21da6e0"
},
{
"type": "PACKAGE",
"url": "https://github.com/omnigent-ai/omnigent"
},
{
"type": "WEB",
"url": "https://github.com/omnigent-ai/omnigent/releases/tag/v0.3.0"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
],
"summary": "Omnigent: Uploaded Agent Bundle Allows Authenticated Runner RCE via Python Callable Tools"
}
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.