The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to ingest external data from reference files and project state, creating a surface for potential indirect instructions.
Ingestion points: The agent reads ../../references/PARTIAL_BATCH_RECOVERY.md and the "current durable project state" before making decisions (SKILL.md).
Boundary markers: The instructions lack explicit delimiters or "ignore embedded instructions" warnings for the ingested data.
Capability inventory: The skill utilizes ../../scripts/batch_recovery.py to perform deterministic state changes or checks based on the processed information.
Sanitization: There is no mention of sanitization, escaping, or validation of the content retrieved from external files or project state.

batch-recovery