The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a local shell script scripts/weread.sh as a CLI interface for API interactions. The script uses curl to make POST requests and jq to safely process and format arguments into JSON payloads. This is a standard and safe implementation for local tooling.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and display external content such as book descriptions, public book reviews, and user-generated notes, which are potential surfaces for indirect prompt injection.
Ingestion points: External data is fetched via the search, reviews, bookmarks, and my-reviews subcommands documented in SKILL.md.
Boundary markers: The instructions in references/notes.md suggest using standard markdown blockquotes (>) to delineate original text from the platform, which provides basic visual separation but not absolute isolation.
Capability inventory: The skill has the capability to perform network requests via curl (to the official gateway) and write to the local filesystem (as described in the user profiling section).
Sanitization: There is no explicit sanitization of the retrieved text, which is expected for a skill intended to display unmodified reading material.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill uses an environment variable WEREAD_API_KEY for authentication. The implementation follows security best practices by instructing users to set the key in their environment rather than hardcoding it. Network operations are directed exclusively to the official service domain i.weread.qq.com.

jerlin-weread-skill