The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs users to install the reedle-cli package via NPM. This is a standard installation for the vendor's provided tool.
[COMMAND_EXECUTION]: The skill relies on the reedle CLI tool via Bash for library management and content extraction. These commands are scoped to the tool's intended functionality.
[CREDENTIALS_UNSAFE]: The skill manages service-specific authentication tokens (Reedle CLI tokens). It describes standard practices for setting, verifying, and storing these tokens in ~/.config/reedle/credentials.json or environment variables.
[PROMPT_INJECTION]: Indirect Prompt Injection Surface: The skill extracts content from untrusted external sources (web pages, YouTube transcripts, Bilibili transcripts) which is then processed by the agent. This presents a surface for indirect prompt injection attacks.
Ingestion points: Untrusted data enters the agent context through the reedle read CLI command and mcp__reedle__reedle_read MCP tools (found in SKILL.md).
Boundary markers: The instructions lack specific guidance on using delimiters or boundary markers to isolate external content from the agent's instructions.
Capability inventory: The agent has the capability to execute shell commands (Bash) and perform state-changing operations on the user's reading library (e.g., reedle_save, reedle_delete_article).
Sanitization: There are no mentioned mechanisms for sanitizing or validating the content extracted from external URLs before the agent processes it.

reedle