The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest large volumes of untrusted data from Discord channels and local cache files. This content is presented to the agent for searching and processing, creating a surface for embedded instructions to influence agent behavior.
Ingestion points: The sync, tail, search, and wiretap commands in SKILL.md read data from external Discord APIs and local Discord Desktop cache files.
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following instructions contained within the mirrored Discord messages.
Capability inventory: The skill utilizes the discrawl CLI which includes subcommands for search, raw SQL execution, and syncing with external Git repositories.
Sanitization: The skill does not describe any sanitization or filtering of the ingested chat data before it is processed by the agent.
[DATA_EXFILTRATION]: The skill targets sensitive local information for its primary purpose. Specifically, the wiretap command accesses local Discord Desktop cache files to import DM history, and the init command handles sensitive Discord Bot Tokens.
[EXTERNAL_DOWNLOADS]: The skill includes a subscribe command that allows the agent to fetch data from user-provided Git repositories. Additionally, the installation instructions point to a Homebrew tap (steipete/tap/discrawl) hosted outside of the primary vendor's GitHub organization.
[COMMAND_EXECUTION]: The sql subcommand enables the execution of arbitrary SQL queries against the local SQLite database containing the mirrored Discord history.

discrawl