agentsclimarketplace

Bulk request scoping

Skill starbridge-ai/plugin.mcp/skills/bulk-request-scoping

Recognize broad, Clay-style bulk-enrichment requests (enrich every row / contact / account) and scope them to a Starbridge-shaped task — narrow, sample, or hand off — before grinding through many slow, approval-gated tool calls.From its SKILL.md

Install
npx -y skills add starbridge-ai/plugin.mcp --skill bulk-request-scoping

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

3.6 KB, as published. Nobody here has run it

Bulk Request Scoping

A guardrail for broad, Clay-style bulk-enrichment asks. Starbridge MCP is built for targeted, per-buyer research and for operating on existing Bridge data — not for bulk table enrichment across hundreds of rows. Slow tools (searchBuyerContacts, getBuyerSummary, web research) chained across many buyers cause approval fatigue and timeouts. This skill catches that shape early and steers it.

When to Use

  • The request implies many buyers × a per-buyer enrichment, especially using slow tools
  • A user asks to enrich, summarize, or find contacts for "all" / "every" / "each" item in a list or bridge
  • Before kicking off a loop that would generate many approval-gated calls

This does not apply to small requests (a handful of buyers) — proceed normally there.

Workflow

  1. Estimate scope. How many buyers × how many per-buyer enrichments, weighted by tool speed (searchBuyerContacts and getBuyerSummary are slow). If the source is a bridge, listBridgeRows (page 1) gives the row count / totalPages.
  2. Decide if it's "large." Treat it as large when many buyers each need a slow per-buyer call — i.e. the plan would produce tens of approval-gated calls or minutes of chained work. (Express the cost in calls/time, not a fixed magic number.) Small jobs: just proceed.
  3. If large, STOP before looping and present options:
    • Narrow — a named subset, or top-N by a stated criterion (e.g. top 10 by score or recency).
    • Sample — run a bounded sample now (e.g. the first 5) so the user sees the shape and cost, then decide whether to continue.
    • Hand off — true bulk enrichment belongs in the Starbridge app (a bridge with enrichment columns) or a Clay-style tool. Explain why, and what Starbridge MCP does well instead (targeted research, operating on existing bridge data).
  4. Proceed only on explicit choice. If the user opts to continue, prefer bulk-capable tools — use getBuyerAttributesBulk (one call) over per-buyer getBuyerAttribute — and batch sensibly.

Tools (used only after scoping)

  • listBridgeRows — to size the job when the source is a bridge.
  • getBuyerAttributesBulk — the bulk path for attribute lookups across many buyers (avoid N single calls).
  • The actual enrichment runs through the relevant skill once scoped: contact-search, buyer-summary, or document-research.

Important

  • This is guidance and stop-conditions, not a producer — once the task is scoped, hand off to the appropriate skill.
  • Don't false-stop small requests; the guard is for many-buyers × slow-per-buyer work.
  • Be explicit about the trade-off (number of calls, approvals, time) so the user can make the call — don't silently grind through a bad plan.

Keep looking

Skills are one crate of 608,475. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.