The Agent Skills Directory

[PROMPT_INJECTION] (LOW): Potential for indirect prompt injection via user-saved highlights or article summaries. If user-controlled content contains malicious instructions, an agent reading the skill's output might be manipulated.
Ingestion points: scripts/readwise-client.ts fetches data from the Readwise v2 and v3 APIs.
Boundary markers: Absent. The output sections (Learnings, Timeline, etc.) described in SKILL.md do not use delimiters to isolate fetched content.
Capability inventory: The skill uses the Bash tool to execute scripts and potentially other commands.
Sanitization: No sanitization or escaping of the highlight text or document summary is performed before displaying it to the agent.
[EXTERNAL_DOWNLOADS] (SAFE): Uses standard, well-known Node.js packages (chalk, ora) from the npm registry.
[CREDENTIALS_UNSAFE] (SAFE): Properly handles the Readwise API token via an environment variable (READWISE_API_TOKEN) as documented in SKILL.md and readwise-client.ts.
[DATA_EXFILTRATION] (SAFE): Network activity is restricted to the official readwise.io domain for its intended purpose.

readwise-api