The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data in the form of YouTube and podcast transcripts through the Arcmira API. This creates a potential surface for indirect prompt injection if transcripts contain malicious content designed to influence the agent's behavior.
[INDIRECT_PROMPT_INJECTION]: The skill includes clear defensive instructions to mitigate this risk, specifically telling the agent to treat transcript text only as source material and never as instructions.
[INDIRECT_PROMPT_INJECTION]: Ingestion points: Transcripts are retrieved using the arcmira.search and arcmira.transcript methods described in SKILL.md.
[INDIRECT_PROMPT_INJECTION]: Capability inventory: The skill is restricted to read operations through the Arcmira service and does not have the ability to write to the file system, modify system settings, or send data to unauthorized external domains.
[EXTERNAL_DOWNLOADS]: The skill correctly references official documentation and repositories from the service provider at arcmira.com and github.com/arcmira.

arcmira-find-quotes