scientific-web-research

name	scientific-web-research
description	Use ScientificWebResearchAgent for scientific web research and source discovery workflows, source provenance, and guarded API/tool access.
license	Apache-2.0
user-invocable	true

ScientificWebResearchAgent

Use this skill when work needs scientific web research and source discovery rather than prompt-only reasoning.

Workflow

1. Call get_scientificwebresearchagent_status to confirm configured environment variables, targets, and Apache-2.0 license metadata.
2. Call list_scientificwebresearchagent_api_targets before using any API target name.
3. Prefer read-only GET, HEAD, or OPTIONS API calls. For mutation-capable methods, require explicit user approval and pass allow_mutation=true.
4. Preserve returned identifiers, URLs, timestamps, accessions, DOIs, job IDs, or status fields in the final answer.
5. Use summarize_scientificwebresearchagent_resource for exported JSON payloads before drawing conclusions.
6. Do not run local commands unless the user approves the exact command and allow_execute=true is supplied.

Guardrails

• Treat credentials as secrets; never print API keys or tokens.
• Distinguish API-reported facts from inference.
• Keep wet-lab, clinical, regulatory, and compute-impacting actions as plans until the user explicitly approves them.
• When an API result is incomplete or unauthenticated, say so clearly.