The Agent Skills Directory

[PROMPT_INJECTION]: No instances of malicious prompt injection, instruction overrides, or safety filter bypasses were detected. The skill uses professional instructional language appropriate for its stated purpose.
[DATA_EXFILTRATION]: No hardcoded credentials, sensitive file access, or unauthorized network exfiltration patterns were identified. Technical paths mentioned, such as /proc/self/status, are used correctly in the context of anti-debugging analysis.
[OBFUSCATION]: The skill documentation is clear and contains no evidence of hidden content, Base64-encoded commands, homoglyphs, or zero-width character obfuscation.
[REMOTE_CODE_EXECUTION]: No remote code execution patterns or downloads of untrusted scripts were found. Command examples are for local diagnostic and reverse engineering tools.
[PRIVILEGE_ESCALATION]: The skill does not include instructions for unauthorized privilege escalation or modification of sensitive system configurations.
[INDIRECT_PROMPT_INJECTION]: While the skill involves processing external challenge artifacts, it maintains a structured methodology that does not expose the agent to unmitigated data-driven instruction overrides.
[DYNAMIC_CONTEXT_INJECTION]: No dynamic context injection patterns using the exclamation mark and backtick syntax were found in the skill or its referenced files.

reverse-ctf