The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and follow instructions from external files located at ../satisfy-your-agent/SKILL.md and ../satisfy-your-agent/references/activities.md. This creates a surface where the agent's behavior is determined by data not contained within the audited skill.
Ingestion points: The skill body in SKILL.md directs the agent to read specific paths outside its own directory.
Boundary markers: There are no delimiters or instructions provided to the agent to treat the external content as untrusted or to ignore embedded instructions.
Capability inventory: The skill explicitly commands the agent to "run" the activity, which may include any capabilities enabled for the agent (shell execution, file manipulation, network access) depending on the external file's content.
Sanitization: No validation or filtering is performed on the ingested content before the agent is instructed to execute it.
[COMMAND_EXECUTION]: The instruction to "run the idle activity" from a referenced markdown file implies that the agent should interpret and execute whatever actions are defined in that external file. This could involve the execution of shell commands or other system-level operations if the referenced file contains such instructions.

sya-idle