The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses coercive and authoritative language such as 'The Iron Law' and 'Violating the letter of this process is violating the spirit' to override the agent's standard behavior and ensure strict adherence to the debugging protocol.
[DATA_EXFILTRATION]: Phase 1, Step 4 directs the agent to 'Log what data enters component' and 'Log what data exits component.' This practice can lead to the accidental exposure of sensitive data, such as credentials or PII, in the agent's log files or terminal output during diagnostic steps.
[INDIRECT_PROMPT_INJECTION]: The skill requires processing untrusted external data, creating an injection vulnerability. Ingestion points: Error messages, stack traces (Phase 1.1), and external reference implementations (Phase 2.2). Boundary markers: Absent; there are no instructions to use delimiters or ignore embedded commands. Capability inventory: Code modification, diagnostic instrumentation (Phase 1.4), and bug fix implementation (Phase 4). Sanitization: Absent; no validation or escaping of external content is specified.
[DYNAMIC_EXECUTION]: The skill involves dynamic code generation and execution for 'diagnostic instrumentation' and bug fixes. This poses a risk if the generated code is influenced by malicious instructions embedded in the external data, such as error logs or stack traces, being analyzed.

debugger