The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use shell commands for content retrieval and processing. Specifically, it recommends using curl to fetch raw HTML and pandoc to convert that HTML to plain text. These are standard tools for this purpose, but they involve direct shell interaction.
[EXTERNAL_DOWNLOADS]: The primary function of this skill is to download external content from arbitrary URLs provided by the user. While this is the stated purpose, it inherently involves fetching data from untrusted remote sources.
[INDIRECT_PROMPT_INJECTION]: By design, the skill fetches the "COMPLETE" and "raw" content of web pages to ensure data fidelity. This creates a surface for indirect prompt injection, as the agent is instructed to ingest and store unvetted content that may contain hidden instructions intended to manipulate the agent's behavior during subsequent tasks.
Ingestion points: Untrusted data enters the context through curl commands fetching user-provided URLs (SKILL.md).
Boundary markers: The skill does not specify the use of delimiters or protective headers when storing or processing the ingested text.
Capability inventory: The skill uses shell execution (curl, pandoc) and file writing (source.md).
Sanitization: There is no mention of sanitizing, escaping, or filtering the retrieved content before it is stored or passed to downstream skills.

flowy-ingest