The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external content such as pull request descriptions, issues, and specifications to build an acceptance contract. This data ingestion creates a surface for indirect prompt injection where an attacker could embed malicious instructions in the requirements. 1. Ingestion points: Identified in SKILL.md as PRs, branches, diffs, and authoritative issues. 2. Boundary markers: The instructions do not specify the use of delimiters or warnings to ignore instructions embedded in the ingested data. 3. Capability inventory: The skill instructions direct the agent to trace production paths and run checks, implying potential tool usage based on the ingested requirements. 4. Sanitization: There are no instructions for sanitizing or escaping the content retrieved from external sources before processing.

acceptance-review