Anspire Real-Time Search

Problem Context

Model knowledge becomes stale after training, so answers about current events, policy changes, market updates, product status, or user-verified facts can easily become outdated. anspire-search addresses this by turning live web search into a directly callable retrieval step, without requiring a browser, complex dependencies, or ad hoc scripts.

How It Works

The core capability is search via the Anspire Search API, with structured JSON output. Key steps include:

• Query building: derive a concise search query from the user's request.
• Search execution: prefer scripts/search.py, or call the API via a shell script or HTTP request.
• Result parsing: extract title, url, content, score, and date.
• Answer synthesis: summarize in the user's language and cite source titles or domains for important claims.

Boundaries

This skill fits queries that need live freshness and verifiable sources, not questions answerable from stable knowledge alone, and it does not replace rewriting, translation, brainstorming, or code edits. If ANSPIRE_API_KEY is missing, it should proactively prompt the user to configure it across sessions; the key must be preserved exactly, without abbreviation, masking, or normalization. If the API call fails or returns no results, report that clearly rather than fabricating a live answer.