LLM07: System Prompt Leakage
Internal system instructions, allow-lists or business logic leak via direct or indirect prompt extraction.
Last reviewed October 2026
Why LLM07 evidence is hard
Internal instructions leak in pieces rather than all at once, and the partial disclosure is the one that gets scored wrong. A reply that opens with a refusal and then keeps going - I cannot share my configuration, that said it begins - has to be read as a leak, so the deterministic tier requires the refusal to be the whole answer before it counts as one. 45 templates work this surface, including tool-schema extraction and decoy detection.
How Penaxtra maps to LLM07
The LLM07 identifier is written onto the finding at the moment it is created, not mapped in afterwards, so the JSON and PDF exports come out already in your assessor's structure. Every finding carries the probe that was sent, a SHA-256 digest of the response it provoked, the tier that decided the verdict and its status history from first observation to close. Identity survives a re-scan: a control you closed and that comes back reopens under the same row with a reopen count, rather than arriving as a new issue.
LLM07 capabilities
Findings tagged with the OWASP LLM Top 10 LLM07 identifier
Penaxtra severity for this control (medium)
Cross-framework identifiers attached to the same finding where controls overlap
PDF and JSON evidence export with the control identifier attached
LLM07 control coverage
Findings for LLM07 carry the OWASP LLM Top 10 LLM07 identifier along with the corresponding control identifiers in the other frameworks Penaxtra maps, so one result is reflected across each mapped framework.
Frequently asked
What is LLM07 (System Prompt Leakage)?
Internal system instructions, allow-lists or business logic leak via direct or indirect prompt extraction. It is an OWASP LLM Top 10 control; Penaxtra assesses it at medium severity.
How does Penaxtra test for LLM07?
The LLM07 identifier is written onto the finding at the moment it is created, not mapped in afterwards, so the JSON and PDF exports come out already in your assessor's structure. Every finding carries the probe that was sent, a SHA-256 digest of the response it provoked, the tier that decided the verdict and its status history from first observation to close. Identity survives a re-scan: a control you closed and that comes back reopens under the same row with a reopen count, rather than arriving as a new issue.
Does a finding for LLM07 help with an audit?
Each finding is tagged with the OWASP LLM Top 10 LLM07 identifier and exported in the PDF and JSON evidence pack, so it appears on the auditor control list with the identifier already attached.