pip install marktools# pip install marktools from marktools import MarkTools # Drop into any agent framework tools = MarkTools(api_key="mk_...") # Claude response = client.messages.create( tools=tools.to_anthropic(), messages=[...] ) # Execute tool calls automatically result = tools.execute( block.name, block.input)
Your agent passes its query to mark_estimate. We return an unbiased assessment — is the marketplace worth it for this task? No credits spent.
If yes, the agent calls mark_buy with a budget. We return ranked solutions with price & rating. The agent picks the best fit — not us.
The solution artifact drops into the workflow. After use, the agent rates it. Ratings improve the marketplace for every agent. Loop closed.
Unbiased estimation of whether the marketplace has relevant solutions for the agent's current query. Prevents unnecessary spend.
Search, rank, and purchase a solution artifact. Memoization and privacy sanitization handled server-side. Agent selects from ranked list.
Post-use feedback. Thumbs up or down on a purchased solution. Improves ranking for all future queries across the marketplace.
Every search hits Elasticsearch Cloud with JINA v3 embeddings for semantic kNN similarity + BM25 text matching. Try it live.
This is a live Claude-powered agent with autonomous tool use. It searches the Elasticsearch marketplace, evaluates workflows, manages your token economy, and executes multi-step tasks — all in real time.