Skip to content

    Type a page, a command such as keryn scan, or a rule id.

    crewai-autogen-agent-safety-v2

    View as Markdown

    8 rules · Keryn Software Licence · rules 0.8.0 · all packs

    How to read a rule
    Severity
    How bad the finding is if it is real: error fails a default gate, warning and info do not.
    Confidence
    P0 exploitable as written; P1 dangerous in context; P2 a heuristic or a posture check: confirm by reading the code.
    False-positive risk
    low: an exact API or flag, so hits are almost always real; medium: the pattern is legitimate in many programs, so check each hit; high: it finds a lead to confirm, so expect benign hits.
    Edge
    The relationship the finding sits on. reaches: content or traffic gets to a thing; calls: one thing invokes another; reads: one thing reads from another; writes: one thing writes to another; sends: data leaves for another party; grants: authority is given or left unchecked; depends_on: something relies on a component it does not verify; trusts: one party believes another without evidence. The boundary it crosses, when there is one: user->prompt (untrusted input reaching a prompt); model->code (model output reaching code, a shell, a query or a page); secret->outside (a secret leaving the trusted side); supply-chain (code, models or prompts pulled in from outside).
    Taxonomy chips
    References for readers who work in OWASP LLM 2025, OWASP Agentic, CWE or NIST AI RMF. They say where a rule belongs, not that a framework is satisfied.

    seceng.agentframework.crewai-agent-dangerous-tools-no-human-gate

    Section titled “seceng.agentframework.crewai-agent-dangerous-tools-no-human-gate”

    info python crewai, autogen

    CrewAI agent appears to use dangerous tools without visible human approval, scope, or allowlist gate.

    Confidence P2: a heuristic or a posture check: confirm by reading the code. False-positive risk medium: the pattern is legitimate in many programs, so check each hit.

    Edge inspected: grants authority is given or left unchecked.

    CWE-862 LLM06 LLM10 ASI02 ASI07 ASI08 GOVERN MANAGE

    seceng.agentframework.crewai-delegation-with-dangerous-tools

    Section titled “seceng.agentframework.crewai-delegation-with-dangerous-tools”

    warning python crewai, autogen

    CrewAI delegation is enabled while dangerous tools or broad tool sets are present. Verify role boundaries and approval flow.

    Confidence P1: dangerous in context. False-positive risk high: it finds a lead to confirm, so expect benign hits.

    Edge inspected: grants authority is given or left unchecked.

    CWE-250 LLM06 ASI02 ASI07 GOVERN MANAGE

    seceng.agentframework.crewai-memory-without-tenant-scope

    Section titled “seceng.agentframework.crewai-memory-without-tenant-scope”

    info python crewai, autogen

    CrewAI memory or knowledge configuration appears shared without visible tenant, user, workspace, or session scoping.

    Confidence P1: dangerous in context. False-positive risk high: it finds a lead to confirm, so expect benign hits.

    Edge inspected: grants authority is given or left unchecked.

    CWE-639 CWE-312 LLM02 LLM08 ASI03 ASI06 ASI02 MAP MANAGE

    seceng.agentframework.crewai-process-without-rate-or-step-limit

    Section titled “seceng.agentframework.crewai-process-without-rate-or-step-limit”

    info python crewai, autogen

    CrewAI process appears to run agents/tasks without visible max iterations, rate limit, timeout, or budget controls.

    Confidence P1: dangerous in context. False-positive risk high: it finds a lead to confirm, so expect benign hits.

    Edge inspected: grants authority is given or left unchecked.

    CWE-770 LLM06 LLM10 ASI02 ASI08 ASI07 MANAGE

    seceng.agentframework.autogen-code-execution-enabled

    Section titled “seceng.agentframework.autogen-code-execution-enabled”

    warning python crewai, autogen, containers

    AutoGen code execution is enabled. Verify Docker sandboxing, working directory isolation, approval, and network restrictions.

    Confidence P2: a heuristic or a posture check: confirm by reading the code. False-positive risk medium: the pattern is legitimate in many programs, so check each hit.

    Edge inspected: calls one thing invokes another; it crosses the model->code boundary (model output reaching code, a shell, a query or a page).

    CWE-94 CWE-78 LLM06 LLM03 ASI05 ASI02 ASI07 GOVERN MANAGE

    seceng.agentframework.autogen-human-input-never-with-tools

    Section titled “seceng.agentframework.autogen-human-input-never-with-tools”

    warning python crewai, autogen

    AutoGen agent appears to run with human_input_mode NEVER while tool/code execution is available.

    Confidence P2: a heuristic or a posture check: confirm by reading the code. False-positive risk medium: the pattern is legitimate in many programs, so check each hit.

    Edge inspected: grants authority is given or left unchecked.

    CWE-862 LLM06 ASI05 ASI02 ASI07 GOVERN MANAGE

    seceng.agentframework.autogen-function-map-dangerous

    Section titled “seceng.agentframework.autogen-function-map-dangerous”

    info python crewai, autogen

    AutoGen function_map/register_function exposes dangerous functions without visible approval, authorization, or scope checks.

    Confidence P2: a heuristic or a posture check: confirm by reading the code. False-positive risk medium: the pattern is legitimate in many programs, so check each hit.

    Edge inspected: grants authority is given or left unchecked.

    CWE-250 LLM06 ASI03 ASI02 ASI07 GOVERN MANAGE

    seceng.agentframework.autogen-shared-workdir-or-history

    Section titled “seceng.agentframework.autogen-shared-workdir-or-history”

    info python crewai, autogen

    AutoGen code execution, chat history, or cache appears to use shared/unscoped work directory or persistent memory.

    Confidence P1: dangerous in context. False-positive risk high: it finds a lead to confirm, so expect benign hits.

    Edge inspected: writes one thing writes to another.

    CWE-639 CWE-312 LLM02 LLM08 ASI05 ASI06 ASI02 MAP MANAGE