The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides tools for retrieving data from external Redis databases, which acts as an ingestion point for potentially untrusted data into the agent's context.
Ingestion points: Data retrieval commands such as redis.get (strings.md), redis.json.get (json.md), index.query (querying.md), and xread (streams.md) ingest content from external database keys.
Boundary markers: The documentation does not include specific boundary markers or instructions to the agent to treat retrieved content as untrusted data rather than instructions.
Capability inventory: The skill enables network communication with the Upstash Redis REST API. AI agents using this skill often have additional capabilities such as file system access or command execution.
Sanitization: The guides focus on standard data access patterns and do not explicitly cover sanitization or validation logic for the retrieved content.
[EXTERNAL_DOWNLOADS]: The skill references and provides installation instructions for official packages and resources from trusted sources.
Evidence: Mentions and provides installation commands for official packages including @upstash/redis, @upstash/ratelimit, redis, and ioredis from the public NPM registry.
Evidence: References official documentation and GitHub repositories within the upstash and upstashdatabase organizations.
[SAFE]: The skill demonstrates secure management of sensitive information and database interactions.
Evidence: Recommends using environment variables for the UPSTASH_REDIS_REST_URL and UPSTASH_REDIS_REST_TOKEN rather than hardcoding credentials.
Evidence: Highlights the Redis.fromEnv() initialization method as a best-practice pattern for initializing the client securely.

upstash-redis-js