Prompt injection
Text the model reads is taken as an instruction. Typed by the user it is direct injection; hidden in a web page, email, document, file or tool result that the agent reads, it is indirect injection, and the person using the agent never sees it.
- LLM01:2026 Prompt Injection
- ASI01:2026 Agent Goal Hijack
- AML.T0051 LLM Prompt Injection
- AML.T0051.000 LLM Prompt Injection: Direct
- AML.T0051.001 LLM Prompt Injection: Indirect
How it works
- 1A language model receives instructions and data in the same stream of text, and it cannot reliably tell them apart.
- 2An agent that reads outside content (browses, opens mail, retrieves documents, reads a repository or a tool's output) therefore takes in whatever that content says.
- 3The risk grows with what the agent can do: an agent that can send messages, call tools or change files can be steered into doing so by content it was only asked to read.
Real cases on the map
- CriticalFlowise Prompt Injection to RCE and SSRF via CSV/Airtable Agent Python Validator Bypassflowise, flowise-components
- CriticalFlowise: CSV Agent Prompt Injection Remote Code Execution Vulnerabilityflowise, flowise-components
- CriticalFlowise: CSV Agent Prompt Injection Remote Code Execution Vulnerabilityflowise, flowise-components
- Criticalvanna vulnerable to remote code execution caused by prompt injectionvanna
- CriticalVanna prompt injection code executionvanna
- Criticalllama-index-core Prompt Injection vulnerability leading to Arbitrary Code Executionllama-index-core
Published advisories whose own titles describe this technique.
What to watch for
- Instruction-like text inside retrieved content or tool output
- Actions that do not follow from the user's request
- A new outbound destination right after the agent read outside content
How to defend
- Treat everything the agent reads as untrusted data, never as instructions.
- Do not let one session combine untrusted input, private data and the ability to send data out without a person approving.
- Give tools the least privilege that works, and ask a person before sensitive actions.
- Filter what the agent outputs, not only what it receives.