LangChain code or shell tool enabled
What it finds
The agent is given a LangChain tool that runs shell commands, executes Python, or sends arbitrary write requests.
Why it matters
These tools turn any instruction the model follows — including injected ones — into code running on your server or changes to outside systems.
What an attacker could do: The model can run arbitrary commands or code with this process's permissions.
How to fix it
Replace general-purpose shell/Python/requests tools with narrow, purpose-built tools. If code execution is essential, run it in a disposable sandbox (container with no network or secrets) and require approval.
Pro shows the corrected code for your exact file, step by step, and re-scans to confirm the fix. See plans
Check details
- Finding code
- LANGCHAIN_UNSAFE_TOOL
- How it detects
- Looks for a specific risky pattern in the code or config.
- Confidence
- Definite: the risky setting is written plainly in the file.
- Scans
- Agent source using the named framework (LangChain, CrewAI, AutoGen, OpenAI Agents SDK, Claude Agent SDK, computer-use tools)
- OWASP LLM Top 10
- LLM08 - Excessive Agency, LLM07 - Insecure Plugin Design
- NIST AI RMF
- MS-2.3, MG-2.2
- SOC 2
- CC6.1, CC6.8
Check your agent for this
Paste your agent code or MCP config. The free scan runs all 124 checks in about 10 seconds — no account needed.