The Agent Skills Directory

[SAFE]: Author Context and Resource Verification. The skill is authored by 'getstream', and all external resources mentioned, such as domains (getstream.io, getstream.dev) and packages (@getstream/*, stream_chat_flutter), are verified vendor-owned assets.
[SAFE]: Secret Management. The skill explicitly forbids hardcoding secrets and provides clear procedures for using the getstream CLI for secure token generation and environment setup, following industry standards for secret isolation.
[SAFE]: Network and Documentation Grounding. The skill uses the WebFetch tool to retrieve official Markdown documentation from trusted getstream.io endpoints and directs the agent to never answer SDK specifics from training data.
[SAFE]: Command Execution. The shell commands and tools referenced, including the Flutter CLI, getstream CLI, jq, and sips, are legitimate development utilities used for project configuration and asset measurement.
[SAFE]: Prompt Injection Mitigation. The skill defines 'non-negotiable rules' and boundary markers that prioritize official live documentation over external input, effectively mitigating the risk of instructions being overridden by malicious third-party data.

stream-flutter