The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external documentation files (.rst, .md) from the docs directory to perform audits and fixes. This creates a surface where instructions or code embedded within the documentation could influence the agent's behavior, particularly during live testing phases.
Ingestion points: documentation files located in docs/ and its subdirectories.
Boundary markers: The skill does not define explicit delimiters or warnings to ignore instructions embedded within the files it audits.
Capability inventory: The skill can execute shell commands, run local Python scripts, and read/write to the file system.
Sanitization: No sanitization of the extracted content is performed before it is passed to the execution runner in Phase C.
[DYNAMIC_EXECUTION]: Phase C of the skill's strategy involves the extraction and execution of code blocks from documentation files using a test runner (scripts/test_doc_code_blocks.py). This dynamic execution of content from files that might be influenced by external contributors or untrusted sources (e.g., in a PR review context) is a significant capability that must be managed carefully.
[COMMAND_EXECUTION]: The skill relies on executing several shell commands and scripts to perform its tasks. These include running embedded Python scripts via HEREDOCs for static scanning (Phases D and E) and invoking external project scripts like scripts/validate_documentation.py. These commands run with the permissions of the agent in the local environment.

devtu-docs-quality