AI Memory Inspector
AI assistants accumulate context across a session — files, selections, and snippets you may not remember sharing. The AI Memory Inspector records that context locally so you can review what was exposed, compare AI responses against the context they were given, and export a report for your own audit trail.
Runs locally in your editor. Secret, PII, prompt-injection, and MCP scanning happen on your machine before anything reaches an AI model.
Start and end memory sessions to capture the context shared with AI, then open the inspector to review it.
Every piece of context routed through the extension is logged locally. Open, export, or clear the ledger — it never leaves your machine unless you export it.
A one-command recap of the most recent session’s exposure, useful right after a heavy AI-assisted refactor.
Compare an AI response against the protected context it was given to spot leaked instructions or unexpected echoes.
Start a memory session
Begin recording at the start of an AI-assisted task.
Work normally
Context shared through the extension is captured in the local ledger as you go.
Inspect and compare
Open the inspector to review exposure and compare responses against the context provided.
Export or clear
Export a local report for your records, or clear the session data entirely.
No security tool is perfect. Here is what this feature does not claim to do, so you can layer defenses appropriately.
Locally, in the extension’s workspace state. Nothing is uploaded unless you explicitly export or enable an opt-in cloud feature.
You can review the local ledger of what was routed through the extension. It cannot attest to channels outside the extension, and we do not claim otherwise.
Exports are redacted of detected secrets and canaries. As with any redaction, review the file before sharing it externally.
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