The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external documentation, including user proposals, referenced specifications, and authoritative contracts. This creates an attack surface where malicious instructions embedded in these documents could influence the agent's architectural deliberations or question generation.
Ingestion points: Phase 1 (Surface & Gap Ingestion) specifies the inspection of user proposals, referenced specifications, and authoritative contracts.
Boundary markers: The instructions do not define explicit delimiters or instructions to ignore embedded commands within the ingested text.
Capability inventory: The skill utilizes non-mutating inspection tools and discovery scripts. While the skill body mandates a read-only state, the agent retains the ability to execute these tools based on findings in the data.
Sanitization: There are no documented procedures for sanitizing or filtering instructions from the external documentation before processing.
[COMMAND_EXECUTION]: The execution protocol explicitly mentions the use of "discovery scripts" to ground inspections in repository documentation. Although the skill mandates that these must be non-mutating, the source and logic of these scripts are not provided within the skill definition, representing an external execution surface.

clarify