The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external Erlang source code and project configuration files (e.g., rebar.config), which represents a potential attack surface for indirect prompt injection.\n
Ingestion points: Erlang modules, process specifications, rebar.config files, and test suites.\n
Boundary markers: The instructions do not define specific delimiters or warnings to ignore instructions embedded within the analyzed project files.\n
Capability inventory: The skill operates with workspace-write permissions and can modify project files, supervision trees, and dependency configurations.\n
Sanitization: No specific validation or escaping mechanisms are described for data read from the project environment.\n- [SAFE]: No malicious intent, obfuscation, or exfiltration patterns were detected. The skill's focus on smallest safe changes and architecture preservation aligns with standard development best practices. The identified injection surface is inherent to coding assistant skills and is mitigated by the skill's emphasis on runtime validation and explicit state compatibility.

erlang-expert