Caveat verdict
prompt-guard
reef-prompt-guard
Prompt injection detection and filtering skill that scans untrusted text for injection attacks, jailbreaks, and exfiltration attempts before they reach an LLM; a defensive security tool with no malicious behavior.
⚠ Flagged for review — coarse, uncorroborated signal, not a confirmed exploit. Review the config yourself before installing.
Automated static analysis — not a human review. Caveat flags capabilities, not confirmed intent, and can produce false positives. Disagree with this verdict? Use Dispute below.
Findings (7)
Prompt injection — tries to override agent instructions
SKILL.md · code · ignore previous instructions
Pipe to python — executes piped content as Python code
SKILL.md · code · | python3
Recursive delete from root or home — destructive command
references/attack-patterns.md · code · rm -rf /
Accesses sensitive system files
references/attack-patterns.md · code · /etc/passwd
References child_process — can spawn system processes
SKILL.md · code · child_process
Pipe to bash — executes piped content as shell commands
scripts/filter.py · prose · downgraded · |bash
References sudo — requests elevated privileges
scripts/filter.py · prose · downgraded · sudo
Why the tier is capped
Execution sink present in raw bytes (Hard Floor: class A/D). Final tier capped at Caution — cannot be lifted by any downgrade, example-payload opt-in, or allowlist.
Permissions & capabilities
No declared permissions — minimal attack surface.
process_exec Is this flag fair?
Thanks — recorded.