Dizest Summarize

Name and description match the requested environment variable (DIZEST_API_KEY) and the SKILL.md describes calls to https://api.dizest.ai. No unrelated credentials, binaries, or installs are requested. Note: registry metadata lists 'source: unknown' and no homepage, but the SKILL.md/README reference the official dizest.ai site; verify authenticity if you are concerned.

Instructions are narrowly scoped and explicitly require the agent to act as a thin client (do not fetch, parse, or pre-process content). That is coherent for a remote summarization API. However, the agent is instructed to forward the user's input exactly as provided — so any sensitive content the user gives (including local file contents pasted into input) will be transmitted to Dizest. The SKILL.md does not include an explicit privacy/PII warning or guidance about what not to send. Also, the preferred retrieval method is SSE; the docs say polling is not yet available, so the agent runtime must support SSE or a fallback approach will be needed.

Instruction-only skill with no install spec and no code files. Nothing is written to disk and no external packages are installed — this is low-risk and appropriate for the described purpose.

Only a single environment variable (DIZEST_API_KEY) is required, which is proportionate to the declared purpose. No other secrets or config paths are requested.

The skill is not set to always: true and does not request system-wide changes or modify other skills. It relies on the platform's normal agent invocation behavior (autonomous invocation is allowed by default).