Skip to main content
The OnePerfectSlice MCP server exposes 9 tools. Each tool maps to one or more API endpoints but is designed for conversational use — flat parameters, composite operations, and error messages that guide the agent to the right next step.

​
list_slices

Discover what analyses are available, or get full details for a specific slice. Example prompts:
  • “What slices do I have access to?”
  • “Show me the details for the high_intent_deals slice”

​
preview_slice_count

Check how many calls match your filters — without starting a run or using AI credits. Example prompts:
  • “How many discovery calls happened last month?”
  • “Preview how many calls match the high_intent_deals slice for Q1”

​
run_slice

Run an AI analysis and get results. Waits up to 30 seconds for completion — if the run takes longer, returns a run ID for polling with get_run. What happens behind the scenes:
  1. Starts an async run via the API
  2. Polls for up to 30 seconds
  3. If completed, automatically fetches evidence and returns everything together
  4. If still running, returns the run ID with a progress message
Example prompts:
  • “Run the high intent deals analysis on last month’s calls”
  • “Analyze discovery calls from the past 2 weeks for the enterprise team”

​
get_run

Retrieve results from a past run, or check progress on a running one. By default it returns the synthesized result without pulling all evidence — drill into evidence only when you need it. Example prompts:
  • “Get the summary for run 12345”
  • “Show me the budget-objection evidence from run 12345”
  • “Check if my last analysis is done yet”

​
list_runs

Browse your team’s run history with optional filtering. Example prompts:
  • “Show me recent completed runs”
  • “What runs are currently active?”
  • “List the last 5 high_intent_deals runs”
  • “Find Jane’s recent runs I can reuse”

​
search_posts

Find AI-generated summaries and scorecards matching your criteria. Returns compact rows by default — each carries a fileId so summary and scorecard rows for the same call can be deduplicated. Use template_type to narrow to one artifact, or detail=true / get_post for full content. Example prompts:
  • “Find scorecards for Jane’s calls last week”
  • “Search for summaries mentioning Acme Corp”
  • “Get the most recent 10 discovery call summaries”

​
get_post

Retrieve a single post with full structured content. Example prompts:
  • “Show me post 456”
  • “Get the full scorecard for that call”

​
search_evidence

Search for evidence quotes across all completed runs. Useful for finding patterns or specific mentions across multiple analyses. Example prompts:
  • “Find evidence of budget objections in discovery calls”
  • “Search for competitive mentions from the last quarter”
  • “Show me intent signals from Jane’s calls last month”
search_evidence filters by element, rep, call type, and date — not by company. To scope evidence to one account, use search_posts (which supports company_domains) or run a slice filtered to that company.

​
get_filter_options

Discover valid values for any filter type. Call this before filtering when you’re not sure what values are available. Available filter types: Example prompts:
  • “What call types are available?”
  • “Search for call owners named Jane”
  • “What CRM deal stages exist?”
  • “Show me available CRM custom fields”