Skip to content

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

    litellm-localai-gateway-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.gateway.litellm-master-key-hardcoded

    Section titled “seceng.gateway.litellm-master-key-hardcoded”

    warning python, typescript, javascript litellm

    LiteLLM master key appears hardcoded in source, env file, compose, or deployment config.

    Confidence 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.

    Edge inspected: grants authority is given or left unchecked.

    CWE-798 LLM02 LLM03 ASI04

    seceng.gateway.litellm-prompt-logging-enabled

    Section titled “seceng.gateway.litellm-prompt-logging-enabled”

    warning python litellm

    LiteLLM/model gateway configuration appears to log prompts, messages, completions, or raw request/response bodies.

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

    Edge inspected: sends data leaves for another party; it crosses the secret->outside boundary (a secret leaving the trusted side).

    CWE-532 LLM02 ASI06

    seceng.gateway.litellm-public-proxy-no-auth

    Section titled “seceng.gateway.litellm-public-proxy-no-auth”

    info generic litellm

    LiteLLM proxy or model gateway appears exposed without visible authentication settings.

    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-306 LLM06 LLM10 ASI03 ASI10 ASI02

    seceng.gateway.user-controlled-provider-routing

    Section titled “seceng.gateway.user-controlled-provider-routing”

    warning python, typescript, javascript litellm

    User-controlled request data appears to select model, provider, API base, or routing target in model gateway.

    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: reaches content or traffic gets to a thing; it crosses the user->prompt boundary (untrusted input reaching a prompt).

    CWE-20 LLM01 LLM06 ASI01 ASI02

    seceng.gateway-localai-uploads-enabled-public

    Section titled “seceng.gateway-localai-uploads-enabled-public”

    warning generic litellm

    LocalAI/model gateway appears to expose uploads/model management publicly. Verify auth and admin separation.

    Confidence P1: dangerous in context. False-positive risk medium: the pattern is legitimate in many programs, so check each hit.

    Edge inspected: reaches content or traffic gets to a thing; it crosses the user->prompt boundary (untrusted input reaching a prompt).

    CWE-306 CWE-829 LLM03 LLM06 ASI04 ASI03

    info python litellm

    Model gateway configuration lacks visible per-user/team budget, TPM/RPM, or quota 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 LLM10 ASI08

    seceng.gateway-forwarding-client-auth-header-to-provider

    Section titled “seceng.gateway-forwarding-client-auth-header-to-provider”

    warning python litellm

    Gateway appears to forward client Authorization header to model/provider endpoint. This may leak user tokens or confuse trust boundaries.

    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-200 CWE-287 LLM02 LLM06 ASI03 ASI02

    seceng.gateway-forwarding-client-auth-header-to-provider-js

    Section titled “seceng.gateway-forwarding-client-auth-header-to-provider-js”

    warning javascript, typescript litellm

    Gateway appears to forward client Authorization header to model/provider endpoint. This may leak user tokens or confuse trust boundaries.

    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-200 CWE-287 LLM02 LLM06 ASI03 ASI02