The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface by instructing the agent to ingest and process external data from existing test files and audit reports to perform automated code consolidation and file management. \n
Ingestion points: The consolidation workflow (described in references/remediation-workflow.md) involves collecting and inventorying behaviors from all test files touching a specific module. \n
Boundary markers: The instructions lack explicit delimiters or warnings to prevent the agent from interpreting comments or strings within test files as instructions during the inventory and rewrite phases. \n
Capability inventory: The agent has the capability to rewrite test files, move files to a quarantine directory, and execute version control commands based on the processed code. \n
Sanitization: No sanitization or validation of the content of the test files is specified before processing or rewriting. \n- [COMMAND_EXECUTION]: The skill directs the agent to perform destructive operations on the local repository for workflow management. \n
Evidence: In references/remediation-workflow.md, the agent is instructed to use git reset --hard HEAD~1 to roll back local changes if a consolidation fails verification.

test-hygiene