Prompt Injection Protection
Prompt injection hides instructions inside content an AI reads — a README, a dependency file, a pasted snippet, or a tool result — to hijack the assistant into leaking data or taking unwanted actions. SoterAI IDE Guard scans the context you are about to share, and the output you get back, for instruction-override and extraction patterns before they can take effect.
Runs locally in your editor. Secret, PII, prompt-injection, and MCP scanning happen on your machine before anything reaches an AI model.
Inspect files and selections for injected instructions (“ignore previous instructions”, hidden directives, and obfuscated variants) before they enter an AI prompt.
Flag suspicious instructions planted in repository files that an assistant might treat as authoritative.
Scan AI output for leaked system instructions, suspicious links, and signs the model was steered off task.
Catch zero-width and combining-mark tricks used to smuggle instructions past a human reviewer.
Scan before prompting
Run a scan of the context you are about to share so hidden instructions are surfaced first.
Review flagged spans
See exactly which text triggered a prompt-injection signal and decide whether to redact or remove it.
Inspect the response
Scan AI output for leaked instructions or suspicious content before you act on it.
Tune thresholds
Adjust sensitivity to balance detection against false positives for your codebase.
No security tool is perfect. Here is what this feature does not claim to do, so you can layer defenses appropriately.
It is when malicious instructions are hidden in content the AI reads — files, web pages, dependency metadata, or tool output — rather than typed by the user. Because assistants often treat that content as trusted, it is a common attack path.
No, and we do not claim to. Detection is heuristic and pattern-based. It meaningfully reduces risk and catches known techniques, but novel attacks can still slip through, so we recommend layered defenses.
Yes. You can scan responses for leaked system instructions, suspicious links, and off-task steering before you rely on them.
Install SoterAI IDE Guard and scan secrets, prompts, MCP tools, and terminal commands locally before they ever reach an AI model.
Install for VS Code