The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill relies on reading and processing existing documentation files within a repository (such as GLOSSARY.md, GLOSSARY-MAP.md, and files under docs/adr/) to detect conflicts and maintain architectural records. Since these files are controlled by users or external contributors, they represent an attack surface where malicious instructions could be embedded to influence the agent's output.
Ingestion points: Processes content from GLOSSARY.md at the repository root, GLOSSARY-MAP.md, and markdown files located in docs/adr/ and docs/research/.
Boundary markers: The instructions do not specify any delimiters or safety prompts to ensure the agent ignores instructions embedded within the processed documentation files.
Capability inventory: The skill is designed to read files, resolve naming conflicts, and write new documentation files to the file system.
Sanitization: There is no evidence of content sanitization or validation to prevent the agent from acting on commands hidden within the ingested documents.

decision-records