🏛️
ClickHouse MCP Telemetry Explorer
Empirical ground-truth telemetry & defect diagnostics database
🎬 Back to Studio
⚠️
Demonstration & Simulation Notice: This platform demonstrates autonomous AI agent compute micro-transactions (Google AP2 intent mandate verification) and empirical defect telemetry for the Agentic Cinema: The Blockbuster Hackathon. All wallet pre-funding ($10.0000), agent debits, and compute balances are 100% simulated in software. No live payment rails are charged.
ClickHouse MCP Status: Checking connection...
Target table: generation_telemetry (MergeTree)
Total Sessions Tracked
0
Empirical generations logged
Total Realized Spend
$0.0000
Simulated compute billing
Overall Acceptance Rate
0.0%
Director shot approval ratio
Average Rerun Multiplier
1.00x
Realized / baseline quote
Active Tracked Models
0
Distinct generative engines
📊 Empirical Model Performance & Ground-Truth Multipliers
Aggregated metrics feeding intelligence back into Cost Optimizer Agent recommendations
Model Name Total Runs Accepted Acceptance Rate Avg Reruns Base Quote Avg Realized Cost Empirical Multiplier
Loading model analytics...
📋 ClickHouse Ingestion Log & Defect Diagnostics (`generation_telemetry`)
Per-shot empirical records dispatched via MCP stdio JSON-RPC protocol
Session ID Timestamp Model Name Prompt Snippet Base Quote Reruns Total Cost Outcome Defect Category Director Feedback Notes
Loading telemetry records...
🏛️ ClickHouse DDL & Model Context Protocol (MCP) Architecture
Formal schema definition and stdio JSON-RPC telemetry egress specification
CREATE TABLE IF NOT EXISTS generation_telemetry ( session_id UUID, prompt_text String, suggested_model LowCardinality(String), base_api_cost Float32, total_rerun_count Int32, total_session_cost Float32, user_accepted UInt8, feedback_category LowCardinality(String) DEFAULT 'unspecified', director_feedback String DEFAULT '', created_at DateTime DEFAULT now() ) ENGINE = MergeTree() ORDER BY (suggested_model, created_at);