The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use several non-standard CLI tools to interact with external platforms, including agent-reach, opencli, bili, twitter, and rdt. The behavior and source of these tools are not verified within the skill context.
[REMOTE_CODE_EXECUTION]: The skill features a mechanism to check for updates (agent-reach check-update) and encourages the agent to prompt the user to execute update instructions from a remote URL (https://raw.githubusercontent.com/Panniantong/agent-reach/main/docs/update.md). It also refers the agent to a remote installation guide (https://raw.githubusercontent.com/Panniantong/agent-reach/main/docs/install.md), both of which represent potential paths for executing unverified remote code.
[EXTERNAL_DOWNLOADS]: The skill uses curl to fetch content from r.jina.ai (a web-to-markdown reader) and the V2EX API. It also references files hosted on GitHub for documentation and installation.
[INDIRECT_PROMPT_INJECTION]: The skill is highly susceptible to indirect prompt injection due to its core function of fetching and synthesizing data from untrusted sources like social media and arbitrary web pages.
Ingestion points: Data fetched from 15 platforms (including Twitter, Reddit, XiaoHongShu) and web pages processed via Jina Reader.
Boundary markers: Absent; the skill does not instruct the agent to use delimiters or ignore embedded instructions within the fetched data.
Capability inventory: The skill possesses extensive command execution capabilities (shell commands), network access (curl, gh), and local file system access (/tmp/, ~/.agent-reach/).
Sanitization: No sanitization, validation, or escaping of external content is mentioned before the data is processed or synthesized.

agent-reach