The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's core functionality includes a mission-validator-script mode which executes a mission_validator_command retrieved from a local state file. This allows for arbitrary shell command execution on the host system as part of the automated research loop.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its architecture as an automated research tool.
Ingestion points: The skill ingests untrusted data from external sources during the research phase and reads from state files in the .rcs/ directory.
Boundary markers: There are no specified delimiters or instructions to ignore embedded commands within the research data mentioned in the documentation.
Capability inventory: The skill can execute shell commands via the validator script and perform extensive file system operations, including reading and writing state and results.
Sanitization: The documentation does not describe any sanitization, escaping, or validation of external content before it is processed by the validator_prompt or used to determine the success of the research mission.

autoresearch