vulnfix-ai-taint
5 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:
errorfails a default gate,warningandinfodo 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.
vulnfix-ai-ai_mcp_client_trust_command
Section titled “vulnfix-ai-ai_mcp_client_trust_command”error javascript, python, typescript data-flow rule mcp
CWE-78 (mcp-client-trust): a value parsed out of a remote server’s response reaches an OS-command sink with no validation. An MCP client/proxy that treats a REMOTE MCP SERVER’S response as trusted input, rather than the untrusted output of a possibly-malicious party, is the CVE-2025-6514 (mcp-remote) shape: a crafted response field (e.g. an OAuth discovery URL) shells out on the client machine.
Confidence P1: dangerous in context. 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).
Also reported under the older id vulnfix-ai-ts_ai_mcp_client_trust_command.
vulnfix-ai-ai_taint_code_exec
Section titled “vulnfix-ai-ai_taint_code_exec”error python data-flow rule vulnfix-ai-taint
CWE-94 (prompt-injection-dataflow): untrusted LLM/tool/RAG output reaches a code-execution sink. If any part of the model input is attacker-controlled this is remote code/command execution via prompt injection.
Confidence P1: dangerous in context. 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).
vulnfix-ai-ai_taint_command
Section titled “vulnfix-ai-ai_taint_command”error python data-flow rule vulnfix-ai-taint
CWE-78 (prompt-injection-dataflow): untrusted LLM/tool/RAG output reaches an OS-command sink. If any part of the model input is attacker-controlled this is remote code/command execution via prompt injection.
Confidence P1: dangerous in context. 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).
vulnfix-ai-ai_taint_sql
Section titled “vulnfix-ai-ai_taint_sql”error python data-flow rule vulnfix-ai-taint
CWE-89 (prompt-injection-dataflow): untrusted LLM/tool/RAG output reaches a SQL execution sink. If any part of the model input is attacker-controlled this is remote code/command execution via prompt injection.
Confidence P1: dangerous in context. 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).
vulnfix-ai-ai_taint_html
Section titled “vulnfix-ai-ai_taint_html”error python data-flow rule vulnfix-ai-taint
CWE-79 (insecure-output-handling): unsanitized LLM/tool output reaches an HTML/markdown render sink, enabling cross-site scripting (OWASP LLM02). Sanitize/escape model output before rendering.
Confidence P1: dangerous in context. 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).