The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines a "Diagnose-and-Fix" pattern that instructs agents to read and act upon full error outputs. This constitutes an attack surface for indirect prompt injection, as malicious instructions could be embedded in error logs from external tools or dependencies to influence the agent's next action.
Ingestion points: Error output from builds, tests, and CLI tools mentioned in the recovery patterns.
Boundary markers: None specified in the instructions.
Capability inventory: The skill itself has no tool capabilities, though it guides agents with execution capabilities (file-write, command execution).
Sanitization: No sanitization logic provided for processing error messages.
[SAFE]: The skill consists of benign documentation outlining standard software engineering practices such as exponential backoff, fallback logic, and graceful degradation. No evidence of obfuscation, hardcoded credentials, or unauthorized network access was found.

error-recovery