The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted meeting transcripts provided by the user, which presents a surface for indirect prompt injection if the transcript contains malicious instructions.
Ingestion points: The agent reads raw .txt transcript files from the user's file system as specified in the workflow and examples (e.g., transcripts/ai-platform-sync/003.txt).
Boundary markers: The instructions do not require the agent to use delimiters or specific safety prompts to ignore instructions embedded within the transcript content when analyzing it.
Capability inventory: The skill leverages Read for data ingestion, and Write and Edit for output generation. The Bash tool is also enabled in the skill's configuration, which could be exploited if an injection successfully compromises the agent's behavior.
Sanitization: There is no evidence of sanitization, validation, or filtering of the content extracted from the transcript before it is processed by the agent.

transcript-notes