MCP prompts and resources

Besides tools, the fypnow MCP server at https://api.fypnow.com/mcp serves prompts and resources through the standard MCP methods prompts/list, prompts/get, resources/list and resources/read. Your client decides how to show them; many list prompts as slash commands. They are the same for every credential, and reading them draws down no plan quota.

Prompts

Each prompt expands into a step-by-step instruction that names the fypnow tools to call and asks the agent to report the figures the tools return.

• weekly_content_plan (argument niche, required): build a week of video ideas for a niche from what is over-performing in it.
• trend_scout (argument niche, optional): scan what is trending now, across the catalogue or inside one niche.
• niche_deep_dive (argument niche, required): study one niche, covering who is winning and which formats, hooks and hashtags repeat.
• creator_deep_dive (argument creator, required): review one tracked creator's results, posting cadence and breakouts.

A prompt only suggests a path. Any step that spends quota, such as generating a script, asks the user first.

Resources

• fypnow://docs/agent-playbook (Agent playbook): which tool answers which question, the metric definitions, which tools spend quota, and how to poll async jobs.
• fypnow://docs/metrics (Metric definitions): the two multiplier definitions and the weighted score with its bands.

Both are served as Markdown. An agent that reads the playbook first learns to rank by multiple or weighted score rather than raw views, to say which multiplier definition it is quoting, and to leave quota-spending tools until the user agrees.

{ "jsonrpc": "2.0", "id": 1, "method": "resources/read",
  "params": { "uri": "fypnow://docs/metrics" } }

See Multiple and weighted score for the formula and Trending videos, hashtags and hooks for the tools the prompts call.