The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill directs the agent to ingest and adhere to content found in GLOSSARY.md and GLOSSARY-MAP.md. These untrusted repository files could be modified by an attacker to include hidden instructions that bypass the intended behavior of the skill.
Ingestion points: GLOSSARY.md and GLOSSARY-MAP.md (referenced in SKILL.md).
Boundary markers: Absent; the instructions do not provide delimiters or warnings to the agent to treat the glossary content as data only rather than instructions.
Capability inventory: No dangerous tools (shell access, file-writing, or network requests) are enabled or utilized in the available files.
Sanitization: Absent; the agent is instructed to use the language from the files directly without a validation or filtering step.

wait-what