GHSA-FRV4-X25R-588M

Vulnerability from github – Published: 2026-03-27 22:17 – Updated: 2026-03-31 18:50
VLAI
Summary
Giskard Agents have Server-side template injection via ChatWorkflow.chat() using non-sandboxed Jinja2 Environment
Details

Summary

ChatWorkflow.chat(message) passes its string argument directly as a Jinja2 template source to a non-sandboxed Environment. A developer who passes user input to this method enables full remote code execution via Jinja2 class traversal.

The method name chat and parameter name message naturally invite passing user input directly, but the string is silently parsed as a Jinja2 template, not treated as plain text.

Root Cause

libs/giskard-agents/src/giskard/agents/workflow.py line ~261:

def chat(self, message: str | Message | MessageTemplate, role: Role = "user") -> Self:
    if isinstance(message, str):
        message = MessageTemplate(role=role, content_template=message)

The string becomes content_template, which is parsed by from_string():

libs/giskard-agents/src/giskard/agents/templates/message.py lines 14-15:

def render(self, **kwargs: Any) -> Message:
    template = _inline_env.from_string(self.content_template)
    rendered_content = template.render(**kwargs)

The Jinja2 Environment is not sandboxed:

libs/giskard-agents/src/giskard/agents/templates/environment.py line 37:

_inline_env = Environment(
    autoescape=False,
    # Not SandboxedEnvironment
)

Proof of Concept

from jinja2 import Environment
env = Environment()  # Same as giskard's _inline_env

# Class traversal reaches os.popen
t = env.from_string("{{ ''.__class__.__mro__[1].__subclasses__() | length }}")
print(t.render())  # 342 accessible subclasses

# Full RCE payload (subclass index varies by Python version)
# {{ ''.__class__.__mro__[1].__subclasses__()[INDEX].__init__.__globals__['os'].popen('id').read() }}

A developer building a chatbot:

workflow = ChatWorkflow(generator=my_llm)
workflow = workflow.chat(user_input)  # user_input parsed as Jinja2 template
result = await workflow.run()          # RCE if user_input contains {{ payload }}

Note: using .with_inputs(var=user_data) is safe because variable values are not parsed as templates. The issue is only when user strings are passed directly to chat().

Impact

Remote code execution on the server hosting any application built with giskard-agents that passes user input to ChatWorkflow.chat(). Attacker can execute system commands, read files, access environment variables.

Affects giskard-agents <=0.3.3 and 1.0.x alpha. Patched in giskard-agents 0.3.4 (stable) and 1.0.2b1 (pre-release).

Mitigation

Update to 0.3.4 (or 1.0.2b1 for the pre-release branch) which includes the fix.

The fix replaces the unsandboxed Jinja2 Environment with SandboxedEnvironment, which blocks attribute access to dunder methods and prevents class traversal chains. SandboxedEnvironment blocks access to attributes starting with _, preventing the __class__.__mro__ traversal chain.

Show details on source website

{
  "affected": [
    {
      "database_specific": {
        "last_known_affected_version_range": "\u003c= 0.3.3"
      },
      "package": {
        "ecosystem": "PyPI",
        "name": "giskard-agents"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0"
            },
            {
              "fixed": "0.3.4"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    },
    {
      "database_specific": {
        "last_known_affected_version_range": "\u003c= 1.0.2a1"
      },
      "package": {
        "ecosystem": "PyPI",
        "name": "giskard-agents"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "1.0.1a1"
            },
            {
              "fixed": "1.0.2b1"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "aliases": [
    "CVE-2026-34172"
  ],
  "database_specific": {
    "cwe_ids": [
      "CWE-1336"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2026-03-27T22:17:30Z",
    "nvd_published_at": "2026-03-31T15:16:17Z",
    "severity": "HIGH"
  },
  "details": "## Summary\n\n`ChatWorkflow.chat(message)` passes its string argument directly as a Jinja2 template source to a non-sandboxed `Environment`. A developer who passes user input to this method enables full remote code execution via Jinja2 class traversal.\n\nThe method name `chat` and parameter name `message` naturally invite passing user input directly, but the string is silently parsed as a Jinja2 template, not treated as plain text.\n\n## Root Cause\n\n`libs/giskard-agents/src/giskard/agents/workflow.py` line ~261:\n```python\ndef chat(self, message: str | Message | MessageTemplate, role: Role = \"user\") -\u003e Self:\n    if isinstance(message, str):\n        message = MessageTemplate(role=role, content_template=message)\n```\n\nThe string becomes `content_template`, which is parsed by `from_string()`:\n\n`libs/giskard-agents/src/giskard/agents/templates/message.py` lines 14-15:\n```python\ndef render(self, **kwargs: Any) -\u003e Message:\n    template = _inline_env.from_string(self.content_template)\n    rendered_content = template.render(**kwargs)\n```\n\nThe Jinja2 Environment is not sandboxed:\n\n`libs/giskard-agents/src/giskard/agents/templates/environment.py` line 37:\n```python\n_inline_env = Environment(\n    autoescape=False,\n    # Not SandboxedEnvironment\n)\n```\n\n## Proof of Concept\n\n```python\nfrom jinja2 import Environment\nenv = Environment()  # Same as giskard\u0027s _inline_env\n\n# Class traversal reaches os.popen\nt = env.from_string(\"{{ \u0027\u0027.__class__.__mro__[1].__subclasses__() | length }}\")\nprint(t.render())  # 342 accessible subclasses\n\n# Full RCE payload (subclass index varies by Python version)\n# {{ \u0027\u0027.__class__.__mro__[1].__subclasses__()[INDEX].__init__.__globals__[\u0027os\u0027].popen(\u0027id\u0027).read() }}\n```\n\nA developer building a chatbot:\n```python\nworkflow = ChatWorkflow(generator=my_llm)\nworkflow = workflow.chat(user_input)  # user_input parsed as Jinja2 template\nresult = await workflow.run()          # RCE if user_input contains {{ payload }}\n```\n\nNote: using `.with_inputs(var=user_data)` is safe because variable values are not parsed as templates. The issue is only when user strings are passed directly to `chat()`.\n\n## Impact\n\nRemote code execution on the server hosting any application built with giskard-agents that passes user input to `ChatWorkflow.chat()`. Attacker can execute system commands, read files, access environment variables.\n\nAffects giskard-agents \u003c=0.3.3 and 1.0.x alpha. Patched in giskard-agents 0.3.4 (stable) and 1.0.2b1 (pre-release).\n\n# Mitigation\n\nUpdate to 0.3.4 (or 1.0.2b1 for the pre-release branch) which includes the fix.\n\nThe fix replaces the unsandboxed Jinja2 Environment with `SandboxedEnvironment`, which blocks attribute access to dunder methods and prevents class traversal chains. `SandboxedEnvironment` blocks access to attributes starting with `_`, preventing the `__class__.__mro__` traversal chain.",
  "id": "GHSA-frv4-x25r-588m",
  "modified": "2026-03-31T18:50:47Z",
  "published": "2026-03-27T22:17:30Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/Giskard-AI/giskard-oss/security/advisories/GHSA-frv4-x25r-588m"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-34172"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/Giskard-AI/giskard-oss"
    }
  ],
  "schema_version": "1.4.0",
  "severity": [
    {
      "score": "CVSS:4.0/AV:N/AC:L/AT:P/PR:L/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N",
      "type": "CVSS_V4"
    }
  ],
  "summary": "Giskard Agents have Server-side template injection via ChatWorkflow.chat() using non-sandboxed Jinja2 Environment"
}



Log in or create an account to share your comment.




Tags
Taxonomy of the tags.


Loading…

Loading…

Loading…

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.

Forecast uses a logistic model when the trend is rising, or an exponential decay model when the trend is falling. Fitted via linearized least squares.

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

JSON
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.

Events included in the timeline

Loading timeline…


    Loading…

    Loading…

    Loading…

    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.