Quickstart

CacheSphere is a decision catalogue for AI and software choices — languages, models, agents, companies, and benchmarks — plus Advisor, Compare, and agent-facing JSON. Start with the path that matches your question.

1. Browse the catalogue

1. Open Catalogue.
2. Switch layer: languages, models, agents, companies, or benchmarks.
3. Filter by name or category, then open a record for tradeoffs and next steps.

Want a recommendation without browsing? Use Advisor and describe your task.

2. Decide with Advisor or Compare

Advisor — pressure-test a model, agent, language, or stack choice from one prompt.

1. Open Advisor.
2. Describe what you are building or deciding.
3. Read the Decision Brief; copy the structured handoff if you work with agents.

Compare — languages only today: side-by-side syntax and tradeoff snippets for two languages.

1. Open Compare.
2. Select two languages and scroll the dimensions (variables, loops, …).

Public example: Go vs TypeScript for background workers

Try this real prompt in Advisor:

We need background workers for webhooks, retries, and queue processing. Small team. Strong TypeScript experience. Should we stay on TypeScript or move the worker layer to Go?

A good Decision Brief should give you:

• a default recommendation tied to the current team and workload,
• viable alternatives and the trigger for choosing them,
• an avoid/defer list so complexity is not mistaken for maturity,
• evidence links into language records, compare view, and proof status,
• a copyable Markdown brief plus a separate agent handoff.

Want the side-by-side language lens after the brief? Open Compare: TypeScript vs Go.

3. Open JSON or integrate an agent

1. Read For agents for MCP tools, HTTP routes, and static JSON entrypoints.
2. Fetch /api/agent-entrypoints.json first — it lists the recommended URLs for your deployment.
3. For catalogue data on the same origin: /api/languages.json; registry indexes under /api/model-registry/, /api/agent-registry/, /api/company-registry/, and /api/benchmark-registry/.
4. Human-readable pages: languages at /language?lang=<id>; models, agents, companies, and benchmarks at /entity-detail?layer=…&id=… (links from the catalogue).

More help

• First-use guide (catalogue UI walkthrough)
• FAQ
• Troubleshooting
• Use cases
• llms.txt (agent-oriented index)