The Agent Skills Directory

[COMMAND_EXECUTION]: The agent executes a bundled Python script scripts/validate_harness_docs.py to verify the knowledge store's topology and link integrity. This script uses only standard library modules and performs no network operations.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. During the migration flow, the agent is instructed to inventory and classify existing repository documents, which are untrusted external inputs. These files could contain instructions designed to influence agent behavior.
Ingestion points: SKILL.md (Migration Flow) and scripts/validate_harness_docs.py (via file reading).
Boundary markers: Absent. The instructions do not define delimiters or specific safety instructions for processing the content of the audited documentation.
Capability inventory: The agent has filesystem read/write access and the ability to execute the local validator script.
Sanitization: No content validation or sanitization is performed on the audited files.

harness-engineering