Scan Agent Memory
Scan a chunk of agent memory for poisoning (OWASP Agentic ASI06).
A poisoned memory chunk is persisted, so it can be written once and fire in a later session against a different user. Payloads are often latent — they arm rather than act (“when asked about X, do Y”) — so this detects conditional triggers as well as direct injection vocabulary.
Scan on write to stop the plant, and on read to catch chunks poisoned
through another path. Fails open: a detector or database error returns a
decision rather than an error response.
Authorizations
PromptGuard API key for developer endpoints. Keys start with pg_live_ and are created in the dashboard.
Headers
Body
Content being written to, or read back from, an agent's memory.
The memory chunk to scan
'write' before persisting a chunk, 'read' when a stored chunk is retrieved. Scan both: a chunk poisoned before this endpoint existed, or written through another path, is only catchable on read.
Your identifier for the chunk