The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources including Slack, email, and Google Drive, which are potentially attacker-controlled.
Ingestion points: The instructions in examples/3p-updates.md, examples/company-newsletter.md, and examples/faq-answers.md direct the agent to use tools to read Slack posts, emails, and shared documents.
Boundary markers: The skill lacks explicit instructions or delimiters to isolate untrusted content or warn the agent to ignore embedded instructions within the ingested data.
Capability inventory: The skill is primarily focused on text generation and drafting. It does not exhibit capabilities for file-system modification or arbitrary command execution.
Sanitization: There is no evidence of content filtering or sanitization for the data retrieved from external communication tools.
[DATA_EXFILTRATION]: The skill instructions encourage the agent to access and summarize sensitive organizational information from internal data sources.
The agent is prompted to search for "critical team members" docs, "leadership updates," and Slack messages in "large channels" or "announce channels" (examples/3p-updates.md, examples/company-newsletter.md).
Accessing these sources is central to the skill's function, but it creates a vector for exposing sensitive internal data within the generated drafts if the output is not carefully reviewed.

internal-comms