The Agent Skills Directory

[PROMPT_INJECTION]: No malicious override instructions or safety bypasses were found. The skill instructions are focused on guiding the agent to act as a 'logging architect' and use the provided reference files to assist users.
[DATA_EXFILTRATION]: The skill actively discourages data exposure. It includes specific patterns for redacting sensitive information (passwords, tokens, PII) from logs. No hardcoded credentials or unauthorized network operations were identified.
[REMOTE_CODE_EXECUTION]: No remote code execution patterns, such as piping curl to bash, were found. The code snippets provided are educational examples of logging implementations using standard libraries.
[EXTERNAL_DOWNLOADS]: The skill references several well-known and trusted Node.js libraries for logging and observability (pino, winston, uuid, prom-client). It also mentions Axiom, which is a well-known logging service. These references are documented as standard implementation patterns.
[COMMAND_EXECUTION]: No dangerous system commands or privilege escalation attempts (like sudo) were detected. The patterns focus on application-level logic.
[OBFUSCATION]: All content is provided in clear text. There is no evidence of Base64, hex encoding, or zero-width character obfuscation.
[INDIRECT_PROMPT_INJECTION]: As a reference-based skill, it processes user-provided code for review. While it lacks explicit boundary markers for untrusted input, its core instructions are centered on enforcing security rules (validations.md) rather than executing arbitrary logic based on that input.
[DYNAMIC_EXECUTION]: No instances of dynamic code generation or execution (eval, exec, etc.) were found in the skill files.

logging-strategies