The Agent Skills Directory

[CREDENTIALS_UNSAFE] (HIGH): The skill requires a DISCORD_USER_TOKEN stored in a .env file. Discord User Tokens grant full access to a user's account and are highly sensitive; storing them in plain text files in the working directory is a security risk.
[PROMPT_INJECTION] (HIGH): The skill is vulnerable to Indirect Prompt Injection (Category 8).
Ingestion points: Untrusted data is fetched from the Discord API, specifically server names, channel names, and DM usernames via discord_list.py.
Boundary markers: There are no delimiters or 'ignore embedded instructions' warnings specified in the output formatting.
Capability inventory: The skill's output is intended to guide the agent's next steps, such as suggesting the use of a discord-sync skill, creating a direct path for malicious server/channel names to influence agent logic.
Sanitization: No sanitization or escaping of the external strings is mentioned, allowing potential instructions embedded in Discord metadata to reach the agent's prompt.
[DATA_EXFILTRATION] (MEDIUM): The skill accesses and exposes a user's entire Discord server and DM structure to the agent's context. While this is the stated purpose, it facilitates the exposure of private communication metadata.
[COMMAND_EXECUTION] (LOW): The skill executes a local Python script ${CLAUDE_PLUGIN_ROOT}/tools/discord_list.py. This is standard functionality but provides the mechanism for processing the aforementioned untrusted data.

discord-list