Semantic Search — Filing Body

Pure vector search over per-chunk full-document embeddings (`filing_embeds`, ~12.4M rows across ~65K filings — each filing sliced into ~190 paragraph-sized chun

liveunknown~5s typical

mcpca-rate-filings

$0.0020

per call

plus $0.001 platform fee per run

Open MCP server; Cracked bills routing only

Provider Upstream docs

$Set up https://cracked-ai-pearl.vercel.app/SKILL.md, then use Cracked to run mcp-ai-serff-ca-rate-filings /search_filing_embeds for me.

Input

Field	Type	Description
query*	string	Natural-language query. Pass the user's question verbatim when you can — short, specific q
topK	integer	Number of top hits to return. Defaults to 10; capped at 50. If filters narrow the candidat
aggregate	boolean	When true (default), collapse to top-K filings by best-chunk similarity. When false, retur
serff	string	Optional SERFF id to scope the chunk search to a single filing (shape PREFIX-IDENTIFIER, e
naic	string	Exact NAIC carrier identifier (5-digit string).
state	string	Two-letter US state code, uppercase. Corpus currently covers CA only.
year	integer	Exact filing year. Mutually exclusive with year_from/year_to.
year_from	integer	Lower bound on filing year, inclusive.
year_to	integer	Upper bound on filing year, inclusive.
product_type	string	Wildcard match on product type. Substring match — "Auto" matches Personal Auto and Commerc
filing_type	string	Wildcard match on filing type ("Rate", "Rule", "Form", etc.). Substring match.
date_from	string	Lower bound on filing date (ISO YYYY-MM-DD).
date_to	string	Upper bound on filing date (ISO YYYY-MM-DD).
predecessor_prefix	string	Bureau / org SERFF prefix ("ISOF", "NCCI", "AAIS", "MSO").

Call it

curl

curl https://cracked-ai-pearl.vercel.app/v1/run \
  -H "Authorization: Bearer ck_live_..." -H "content-type: application/json" \
  -d '{"provider":"mcp-ai-serff-ca-rate-filings","endpoint":"/search_filing_embeds","input":{"query":"..."}}'

cli

npx cracked-ai run -p mcp-ai-serff-ca-rate-filings -e /search_filing_embeds -i '{"query":"..."}'

mcp

run_tool({ provider: "mcp-ai-serff-ca-rate-filings", endpoint: "/search_filing_embeds", input: {"query":"..."} })

Try it

Runs against your signed-in workspace balance. Sign in if you have not.