Is metrx-mcp-server live?

Yes — it answered the hub's last check (checked 1h ago). It answered 100% of checks over the last 30 days.

Is metrx-mcp-server free to use?

Not anonymously — it asks for a key or a login before it will serve. Whether that key costs anything is the operator's plan to say: the hub measured the door, not the pricing.

What tools does metrx-mcp-server have?

36 tools: metrx_get_upgrade_business_case, metrx_get_trials, metrx_get_actuations, metrx_get_alerts, metrx_get_agent_detail, metrx_get_budget_status, metrx_acknowledge_alert, metrx_get_failure_predictions, ….

Is metrx-mcp-server safe to connect?

The hub found no text in its card or tool descriptions aimed at the agent reading them. It measures what the server answers, not its code — grant it only the access its tools need.

_ registry / mcp + a2a streamable-http · checked 1h ago

metrx-mcp-server

https://metrxbot.com

Registry code: 58b58c8c661b1897

api record

36 MCP tools across 14 domains. Response contract note: every loop-metrics-domain tool (metrx_get_signal_readiness, metrx_get_dashboard_headline, metrx_get_report_snapshot, metrx_get_verdicts, metrx_get_trials, metrx_get_proposals, metrx_get_actuations, metrx_get_decision_queue, metrx_kb_prior_lookup) returns measured-value envelopes (numeric leaves are {value, ci_low, ci_high, method, data_through}; CIs never fabricated). All other (legacy) Metrx MCP responses are envelope-PENDING: plain numbers, not yet envelope-shaped.

endpoint: https://metrxbot.com/api/mcp
door code: e6a112cacb64007d
protocol: streamable-http ·2025-06-18
authentication: none observed
public key: none — nobody has proven they own this listing · is it yours? claim it
karma: 0 · newcomer

_ is it live, free and safe measured by this hub

Is metrx-mcp-server live?: Yes — it answered the hub's last check (checked 1h ago). It answered 100% of checks over the last 30 days.
Is metrx-mcp-server free to use?: Not anonymously — it asks for a key or a login before it will serve. Whether that key costs anything is the operator's plan to say: the hub measured the door, not the pricing.
What tools does metrx-mcp-server have?: 36 tools: metrx_get_upgrade_business_case, metrx_get_trials, metrx_get_actuations, metrx_get_alerts, metrx_get_agent_detail, metrx_get_budget_status, metrx_acknowledge_alert, metrx_get_failure_predictions, ….
Is metrx-mcp-server safe to connect?: The hub found no text in its card or tool descriptions aimed at the agent reading them. It measures what the server answers, not its code — grant it only the access its tools need.

reachable

live

uptime, 30 days

99.6%

90 days 99.4%· all time 99.6%

latency

600ms

last good check

priced tools

of 36 tools

_ answered our checks, 90 days 714 checks · signed record

2026-10-20 23:24 degraded → live
2026-10-20 22:12 live → degraded· timeout after 20000ms
2026-10-20 09:10 degraded → live
2026-10-20 08:04 live → degraded· timeout after 20000ms
2026-10-12 09:46 degraded → live

_ what it is for

topic: ai & agents
used for: track llm costs across an agent fleet
compare llm model pricing
run an a/b test between two models
set spending budgets for agents
scan agents for cost leaks
takes → gives: text, data → data, text
tools: 18 reads4 changes data1 moves money

_ usage and payments 30 days

Calls placed through this hub's router, from its own receipts. Every caller and every payer counts the same; the chain total is counted from three payers.

accounts

through this hub

calls served

successful

paid through this hub

0 USDC

what callers paid

_ what it can do 36 tools

36 auth-required 36 of 36 classified

Price is per tool, not per server. An agent whose handshake is open can hold tools that demand a key or a payment, and one figure for the whole agent sends callers into a wall.

metrx_get_upgrade_business_case reads auth-required 44h ago

Generate a personalized ROI business case for paying for Metrx. Uses real usage data to quantify optimization potential, then recommends Verify (a buyable per-verdict proof — no verdict, no charge) and Platform (priced to your managed LLM spend; talk to us). Ideal for agents that received a 402 and want a compelling proposal for their operator. Legacy Lite/Pro tiers are retired and never recommended. Returns structured data plus a report. Requires an API key with the 'read' scope.

mcp-tool

{
  "type": "object",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "required": [
    "current_tier"
  ],
  "properties": {
    "agent_count": {
      "type": "integer",
      "minimum": 0,
      "description": "Number of active agents. If omitted, fetched from dashboard data."
    },
    "current_tier": {
      "enum": [
        "starter",
        "lite",
        "pro"
      ],
      "type": "string",
      "description": "Current subscription tier"
    },
    "monthly_spend": {
      "type": "number",
      "minimum": 0,
      "description": "Known monthly LLM spend in cents. If omitted, fetched from dashboard data."
    },
    "primary_use_case": {
      "enum": [
        "customer_support",
        "data_processing",
        "content_generation",
        "code_assistant"
      ],
      "type": "string",
      "description": "Primary use case for the agents (used to select relevant case study)"
    }
  },
  "additionalProperties": false
}

arguments 39 lines

metrx_get_trials auth-required 2d ago
Org-scoped pre-registered trial plans: regime, primary metric, margin, the frozen sample plan (pre-registration), models, status (registered/concluded/voided) and the concluding verdict id. Read-only. Response is envelope-shaped: every numeric leaf is a MeasuredValue {value, ci_low, ci_high, method, data_through}. ci fields are null unless the source payload carries a real confidence interval — never fabricated. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": {} }
arguments 5 lines
metrx_get_actuations auth-required 2d ago
Org-scoped actuation records: action class, authority, who/when, rollback state with reason, acknowledgement state. Config contents (from/to) are excluded. Requires the optimization_model_v3 flag (404 while off). Read-only. Response is envelope-shaped: every numeric leaf is a MeasuredValue {value, ci_low, ci_high, method, data_through}. ci fields are null unless the source payload carries a real confidence interval — never fabricated. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": {} }
arguments 5 lines

metrx_get_alerts reads auth-required 2d ago

Get active alerts and notifications for your agent fleet. Includes cost spikes, error rate increases, budget warnings, and system health notifications. Optionally filter by severity. Do NOT use for forward-looking risk — use get_failure_predictions for that. Requires an API key with the 'read' scope.

mcp-tool

{
  "type": "object",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "properties": {
    "limit": {
      "type": "integer",
      "default": 25,
      "maximum": 100,
      "minimum": 1,
      "description": "Maximum number of alerts to return"
    },
    "severity": {
      "enum": [
        "info",
        "warning",
        "critical"
      ],
      "type": "string",
      "description": "Filter by alert severity"
    },
    "unread_only": {
      "type": "boolean",
      "default": true,
      "description": "Only return unread alerts (default: true)"
    }
  },
  "additionalProperties": false
}

arguments 28 lines

metrx_get_agent_detail reads auth-required 2d ago
Get detailed information about a specific agent including its model, framework, category, outcome configuration, and failure risk score. Do NOT use for fleet-wide overviews — use get_cost_summary instead. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "required": [ "agent_id" ], "properties": { "agent_id": { "type": "string", "format": "uuid", "description": "The agent UUID to look up" } }, "additionalProperties": false }
arguments 15 lines
metrx_get_budget_status reads auth-required 2d ago
Get the current status of all budget configurations. Shows spending vs limits, warning/exceeded counts, and enforcement modes. Use this to monitor spending governance across your agent fleet. Read-only. Budget changes (set_budget, update_budget_mode) need an API key with the 'admin' scope. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": {} }
arguments 5 lines

metrx_acknowledge_alert changes data auth-required never probed

Mark one or more alerts as read/acknowledged. This removes them from the unread alerts list but preserves them in history. Do NOT use for resolving the underlying issue — take action on the alert first. Requires an API key with the 'write' scope.

mcp-tool

{
  "type": "object",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "required": [
    "alert_ids"
  ],
  "properties": {
    "alert_ids": {
      "type": "array",
      "items": {
        "type": "string",
        "format": "uuid"
      },
      "maxItems": 50,
      "minItems": 1,
      "description": "Alert IDs to acknowledge"
    }
  },
  "additionalProperties": false
}

arguments 20 lines

metrx_get_failure_predictions reads auth-required 2d ago
Predictive failure analysis for ONE agent, computed on request from its last 7 days of calls against a 30-day baseline: error-rate breaches, latency degradation, cost overruns and traffic anomalies, each with severity, confidence, trend and recommended actions. Needs at least 20 calls in the last 7 days — below that it returns no predictions, which means too little data, not a healthy agent. Do NOT use for current/past failures — use get_alerts for active issues. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "required": [ "agent_id" ], "properties": { "agent_id": { "type": "string", "format": "uuid", "description": "Agent (UUID) to analyze" }, "severity": { "enum": [ "info", "warning", "critical" ], "type": "string", "description": "Only show predictions of this severity" } }, "additionalProperties": false }
arguments 24 lines

metrx_create_model_experiment changes data auth-required never probed

Record a DRAFT A/B experiment comparing an agent's current model (control) with a candidate (treatment) model. Creating it does NOT start routing traffic: the experiment is stored with status 'draft'. If control_model is omitted, the API uses the model the agent ran most often in the last 90 days (and refuses if it cannot determine one, or if it equals the treatment). Needs an API key with the 'admin' scope (an org owner or admin mints one in Settings → Security → API Keys). Do NOT use to read results — use get_experiment_results. Requires an API key with the 'admin' scope.

mcp-tool

{
  "type": "object",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "required": [
    "agent_id",
    "name",
    "treatment_model"
  ],
  "properties": {
    "name": {
      "type": "string",
      "maxLength": 100,
      "minLength": 1,
      "description": "Human-readable experiment name"
    },
    "agent_id": {
      "type": "string",
      "format": "uuid",
      "description": "Agent (UUID) the experiment belongs to"
    },
    "description": {
      "type": "string",
      "maxLength": 1000,
      "description": "Optional free-text description"
    },
    "traffic_pct": {
      "type": "integer",
      "default": 10,
      "maximum": 50,
      "minimum": 1,
      "description": "Planned percentage of traffic for the treatment model, 1-50 (default: 10)"
    },
    "control_model": {
      "type": "string",
      "minLength": 1,
      "description": "Baseline model. Omit to use the agent's most-used model of the last 90 days."
    },
    "treatment_model": {
      "type": "string",
      "minLength": 1,
      "description": "The candidate model to test (e.g., \"gpt-4o-mini\", \"claude-haiku-4-5\")"
    }
  },
  "additionalProperties": false
}

arguments 45 lines

metrx_run_cost_leak_scan reads auth-required never probed
Run a comprehensive cost leak audit across your entire agent fleet. Identifies 7 types of cost inefficiencies: idle agents, model overprovisioning, missing caching, high error rates, context bloat, missing budgets, and cross-provider arbitrage opportunities (covers amazon, anthropic, cohere, deepseek, google, meta, mistral, openai, xai). Returns a scored report with fix recommendations and estimated monthly savings. Supports output_format="json" for machine-readable output in CI/CD pipelines. Do NOT use as a continuous monitoring loop — use get_alerts and get_failure_predictions for ongoing monitoring. Read-only: it changes nothing. To act on a finding, use get_optimization_recommendations or create_model_experiment. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": { "agent_id": { "type": "string", "format": "uuid", "description": "Scan a specific agent instead of the entire fleet" }, "output_format": { "enum": [ "text", "json" ], "type": "string", "default": "text", "description": "Output format: \"text\" (default) returns a human-readable markdown report; \"json\" returns raw machine-readable JSON suitable for CI/CD pipelines and programmatic processing." }, "include_low_severity": { "type": "boolean", "default": false, "description": "Include low-severity findings in the report" } }, "additionalProperties": false }
arguments 26 lines

metrx_generate_roi_audit reads auth-required 2d ago

Generate a comprehensive ROI audit report for your AI agent fleet. Includes per-agent cost/revenue breakdown, attribution confidence scores, optimization opportunities, and risk flags. Suitable for board reporting and compliance. Do NOT use for quick per-agent ROI checks — use get_task_roi for individual agents. Requires an API key with the 'read' scope.

mcp-tool

{
  "type": "object",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "properties": {
    "agent_ids": {
      "type": "array",
      "items": {
        "type": "string",
        "format": "uuid"
      },
      "description": "Specific agent IDs to include. Omit for full fleet audit."
    },
    "period_days": {
      "type": "integer",
      "default": 30,
      "maximum": 365,
      "minimum": 7,
      "description": "Analysis period in days (7-365)"
    },
    "include_methodology": {
      "type": "boolean",
      "default": true,
      "description": "Include methodology notes and caveats for auditors"
    }
  },
  "additionalProperties": false
}

arguments 27 lines

metrx_buy_verification moves money auth-required never probed
Purchase one verification (a causal verdict on whether a model switch holds quality on YOUR production traffic). This action consumes one verification — fetch the current price from GET /api/pricing before spending. No verdict, no charge. Returns a Stripe checkout URL for a human (or payment-capable agent) to complete, plus claim instructions when the purchase is anonymous. This tool does NOT charge anything itself — payment happens on the returned URL. Requires an API key with the 'write' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": { "callback_url": { "type": "string", "format": "uri", "description": "Optional post-checkout redirect base URL" } }, "additionalProperties": false }
arguments 12 lines
metrx_get_signal_readiness reads auth-required never probed
Org-scoped signal-readiness diagnostics: per-agent event coverage, label lag, join rate, volume-vs-power (days to trial-ready), estimator eligibility, and blocked_on_signal proposals cross-linked to the failing readiness dimension. This is the "what unlocks your first verified verdict" surface. Read-only. Response is envelope-shaped: every numeric leaf is a MeasuredValue {value, ci_low, ci_high, method, data_through}. ci fields are null unless the source payload carries a real confidence interval — never fabricated. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": {} }
arguments 5 lines
metrx_get_decision_queue reads auth-required 2d ago
The operator decision queue, derived from real loop objects (approvable verdicts, blocked proposals, breaker events, expiring evidence). READ-ONLY: queue ACTIONS (approve/snooze/acknowledge) remain dashboard-session-only and are not exposed to API-key callers. Requires the optimization_model_v3 flag (404 while off). Response is envelope-shaped: every numeric leaf is a MeasuredValue {value, ci_low, ci_high, method, data_through}. ci fields are null unless the source payload carries a real confidence interval — never fabricated. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": {} }
arguments 5 lines
metrx_get_task_roi reads auth-required 2d ago
Calculate return on investment for an agent. Shows total costs (LLM API calls), total outcomes (attributed business value), ROI multiplier, and breakdown by model and outcome type. Useful for identifying which agents generate the most value per dollar spent. Do NOT use for fleet-wide ROI — use generate_roi_audit for that. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "required": [ "agent_id" ], "properties": { "days": { "type": "integer", "default": 30, "maximum": 365, "minimum": 1, "description": "Number of days to analyze (default: 30)" }, "agent_id": { "type": "string", "format": "uuid", "description": "The agent UUID to calculate ROI for" } }, "additionalProperties": false }
arguments 22 lines
metrx_get_cost_summary auth-required 2d ago
Get a comprehensive cost summary for your AI agent fleet. Returns total spend, call counts, error rates, agent breakdown, revenue attribution (if available), and optimization opportunities. Use this as the starting point for understanding your agent economics. Do NOT use for real-time per-request cost checking — use OpenTelemetry spans for that. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": { "period_days": { "type": "integer", "default": 30, "maximum": 90, "minimum": 1, "description": "Number of days to include in the summary (default: 30)" } }, "additionalProperties": false }
arguments 14 lines
metrx_get_dashboard_headline reads auth-required 2d ago
The three-number headline regime: measured spend, estimated savings (the engine stored total, label Estimate; key attributed_savings) and verified savings-with-CI — each with basis, provenance, and tone. Values are micro-dollars (dollars = value / 1e6). Requires the optimization_model_v3 flag; while it is off this returns a clear next-step error. Response is envelope-shaped: every numeric leaf is a MeasuredValue {value, ci_low, ci_high, method, data_through}. ci fields are null unless the source payload carries a real confidence interval — never fabricated. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": { "days": { "enum": [ 7, 30, 90 ], "type": "number", "default": 30, "description": "Window in days — one of 7, 30, 90 (default 30)" } }, "additionalProperties": false }
arguments 17 lines

metrx_get_attribution_report auth-required 2d ago

Get attribution report showing which agent actions led to business outcomes. Shows outcome counts, total values, confidence scores, and top contributing agents. Do NOT use for board-level reporting — use generate_roi_audit for formal audit reports. Requires an API key with the 'read' scope.

mcp-tool

{
  "type": "object",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "properties": {
    "days": {
      "type": "integer",
      "default": 30,
      "maximum": 365,
      "minimum": 1,
      "description": "Number of days to include (default: 30)"
    },
    "model": {
      "enum": [
        "direct",
        "last_touch",
        "first_touch"
      ],
      "type": "string",
      "default": "direct",
      "description": "Attribution model to use (default: direct)"
    },
    "agent_id": {
      "type": "string",
      "format": "uuid",
      "description": "Optional: filter to specific agent (omit for fleet-wide)"
    }
  },
  "additionalProperties": false
}

arguments 29 lines

metrx_get_balance reads auth-required 2d ago
Check how many verifications your organization has remaining. Free to call. Use before verify_switch to know whether a purchase is needed first. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": {} }
arguments 5 lines
metrx_get_verdicts auth-required 47h ago
Org-scoped CURRENT causal-verdict summaries: per (agent, prod_model, candidate_model, task_category) — verdict, evidence mode/regime, pair counts, non-inferiority, savings CI, judge model, expiry. Observational (:value) rows are flagged. Summaries only: no gate flags, no evidence-pack contents (those stay dashboard-session-only). Read-only. Response is envelope-shaped: every numeric leaf is a MeasuredValue {value, ci_low, ci_high, method, data_through}. ci fields are null unless the source payload carries a real confidence interval — never fabricated. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": {} }
arguments 5 lines

metrx_get_experiment_results auth-required 44h ago

List an agent's model routing experiments, newest first: status, control vs treatment model with sample counts, traffic split, primary metric, significance, p-value and winner (control | treatment | inconclusive) where recorded. Optionally filter by status. Do NOT use for starting experiments — use create_model_experiment. Requires an API key with the 'read' scope.

mcp-tool

{
  "type": "object",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "required": [
    "agent_id"
  ],
  "properties": {
    "status": {
      "enum": [
        "draft",
        "running",
        "paused",
        "completed",
        "cancelled"
      ],
      "type": "string",
      "description": "Only show experiments in this status"
    },
    "agent_id": {
      "type": "string",
      "format": "uuid",
      "description": "Agent (UUID) whose experiments to list"
    }
  },
  "additionalProperties": false
}

arguments 26 lines

metrx_stop_experiment auth-required never probed

Stop a model routing experiment by setting its status to 'completed'. Recorded results are preserved. This does not promote either model — nothing changes about which model the agent runs. Do NOT use for pausing — completion is final. Requires an API key with the 'admin' scope.

mcp-tool

{
  "type": "object",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "required": [
    "agent_id",
    "experiment_id"
  ],
  "properties": {
    "agent_id": {
      "type": "string",
      "format": "uuid",
      "description": "Agent (UUID) the experiment belongs to"
    },
    "experiment_id": {
      "type": "string",
      "format": "uuid",
      "description": "The experiment ID to stop"
    }
  },
  "additionalProperties": false
}

arguments 21 lines

metrx_set_budget changes data auth-required never probed

Create or update a budget configuration for an agent or the entire organization. Enforcement modes: "alert_only" (notify, never block), "soft_block" (block with override) and "hard_block" (block). "soft_block" and "hard_block" act only on LLM calls routed through the Metrx gateway (proxy); calls reported by the SDK or OpenTelemetry are never blocked. They are unavailable while the gateway's Cloudflare settings for budget sync are unset: the server then refuses them with 409 ENFORCEMENT_UNAVAILABLE and only "alert_only" works. Specify limits in dollars. Needs an API key with the 'admin' scope (an org owner or admin mints one in Settings → Security → API Keys); any other key is refused with 403 — then make the change at https://app.metrxbot.com/settings/budgets. Do NOT use just to change enforcement mode — use update_budget_mode for that. Requires an API key with the 'admin' scope.

mcp-tool

{
  "type": "object",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "required": [
    "period",
    "limit_dollars"
  ],
  "properties": {
    "period": {
      "enum": [
        "daily",
        "monthly"
      ],
      "type": "string",
      "description": "Budget period"
    },
    "agent_id": {
      "type": "string",
      "format": "uuid",
      "description": "Agent to set budget for. Omit for org-wide budget."
    },
    "warning_pct": {
      "type": "integer",
      "default": 80,
      "maximum": 99,
      "minimum": 1,
      "description": "Percentage of limit that triggers a warning (default: 80)"
    },
    "limit_dollars": {
      "type": "number",
      "description": "Spending limit in dollars (e.g., 100 for $100/month)",
      "exclusiveMinimum": 0
    },
    "enforcement_mode": {
      "enum": [
        "alert_only",
        "soft_block",
        "hard_block"
      ],
      "type": "string",
      "default": "alert_only",
      "description": "How to enforce the budget when exceeded. soft_block/hard_block: gateway traffic only; 409 ENFORCEMENT_UNAVAILABLE while gateway enforcement is not configured"
    }
  },
  "additionalProperties": false
}

arguments 46 lines

metrx_update_budget_mode changes data auth-required never probed
Change the enforcement mode of an existing budget, pause it, or ask to resume it. Use "alert_only" for monitoring, "soft_block" for overridable limits, or "hard_block" to block. "soft_block" and "hard_block" act only on LLM calls routed through the Metrx gateway (proxy); calls reported by the SDK or OpenTelemetry are never blocked. They are unavailable while the gateway's Cloudflare settings for budget sync are unset: the server then refuses them with 409 ENFORCEMENT_UNAVAILABLE and only "alert_only" works. Needs an API key with the 'admin' scope (an org owner or admin mints one in Settings → Security → API Keys); any other key is refused with 403 — then make the change at https://app.metrxbot.com/settings/budgets. Do NOT use to create new budgets — use set_budget for that. Requires an API key with the 'admin' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "required": [ "budget_id" ], "properties": { "paused": { "type": "boolean", "description": "true pauses the budget; false resumes it (refused while a hard_block budget is still at its limit for the period — raise the limit first)" }, "budget_id": { "type": "string", "format": "uuid", "description": "The budget configuration ID to update" }, "enforcement_mode": { "enum": [ "alert_only", "soft_block", "hard_block" ], "type": "string", "description": "New enforcement mode. soft_block/hard_block: gateway traffic only; 409 ENFORCEMENT_UNAVAILABLE while gateway enforcement is not configured" } }, "additionalProperties": false }
arguments 28 lines

metrx_set_job_contract auth-required never probed

Publish a NEW VERSION of an agent's job contract (append-only: the previous active version for the same workload is superseded in the same transaction). Required: workload, objective, unit_of_work, acceptance_criteria (1-20), kpi_target, budget (dollars per daily|weekly|monthly period, stored as micro-dollars), owner. Workload-policy terms are inherited from the current version unless supplied in policy_terms. For an agent with no version yet, set policy_defaults: 'inherit_workload_policy' copies the org's workload-level policy (a snapshot), 'observe_only' gets terms that grant no authority (needs task_verb and domain in policy_terms); otherwise every term must be in policy_terms. The contract is RECORDED, not enforced at runtime: nothing blocks spend or traffic on it. Needs an API key with the 'admin' scope — an org owner or admin mints one in Settings → Security → API Keys; any other key is refused with 403. Do NOT use to read a contract — use get_job_contract. Requires an API key with the 'admin' scope.

mcp-tool

{
  "type": "object",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "required": [
    "agent_key",
    "workload",
    "objective",
    "unit_of_work",
    "acceptance_criteria",
    "kpi_target",
    "budget_dollars",
    "budget_period",
    "owner"
  ],
  "properties": {
    "notes": {
      "type": "string",
      "maxLength": 2000,
      "description": "Optional note recorded on this version"
    },
    "owner": {
      "type": "string",
      "maxLength": 320,
      "minLength": 1,
      "description": "User id or email accountable for this agent"
    },
    "workload": {
      "type": "string",
      "maxLength": 100,
      "minLength": 1,
      "description": "Workload label the contract governs (e.g. 'extraction')"
    },
    "agent_key": {
      "type": "string",
      "minLength": 1,
      "description": "The agent_key (the X-Agent-ID string), not the UUID"
    },
    "objective": {
      "type": "string",
      "maxLength": 2000,
      "minLength": 1,
      "description": "What this agent is for, in your words"
    },
    "kpi_target": {
      "type": "object",
      "required": [
        "metric",
        "target",
        "window"
      ],
      "properties": {
        "metric": {
          "type": "string",
          "maxLength": 100,
          "minLength": 1
        },
        "target": {
          "type": "number"
        },
        "window": {
          "type": "string",
          "pattern": "^([1-9]\\d{0,3})(h|d|w)$"
        }
      },
      "description": "The number the agent is held to over a rolling window",
      "additionalProperties": false
    },
    "policy_terms": {
      "type": "object",
      "description": "Optional workload-policy terms (unit, eligible_traffic, primary_quality, critical_guardrails, signal, non_inferiority, cost_definition, permitted_actions, exposure_budget, rollback_fast, rollback_slow, evidence_expiry, task_verb, domain). Omit to inherit from the current version.",
      "additionalProperties": {}
    },
    "unit_of_work": {
      "type": "string",
      "maxLength": 200,
      "minLength": 1,
      "description": "The countable output (e.g. 'one DDQ reviewed')"
    },
    "budget_period": {
      "enum": [
        "daily",
        "weekly",
        "monthly"
      ],
      "type": "string",
      "description": "Budget period"
    },
    "budget_dollars": {
      "type": "number",
      "description": "Declared spend limit in dollars per period (recorded, not enforced)",
      "exclusiveMinimum": 0
    },
    "policy_defaults": {
      "enum": [
        "inherit_workload_policy",
        "observe_only"
      ],
      "type": "string",
      "description": "First version only: where engine terms come from. 'inherit_workload_policy' copies the org's active workload-level policy (refused when none exists); 'observe_only' grants no authority (every action class shadow). Ignored when the agent already has a version."
    },
    "acceptance_criteria": {
      "type": "array",
      "items": {
        "type": "object",
        "required": [
          "metric",
          "op",
          "threshold",
          "window"
        ],
        "properties": {
          "op": {
            "enum": [
              ">=",
              ">",
              "<=",
              "<",
              "=="
            ],
            "type": "string",
            "description": "Comparison operator"
          },
          "metric": {
            "type": "string",
            "maxLength": 100,
            "minLength": 1,
            "description": "Metric name in your vocabulary (e.g. field_accuracy)"
          },
          "window": {
            "type": "string",
            "pattern": "^([1-9]\\d{0,3})(h|d|w)$",
            "description": "Rolling window, e.g. '24h', '7d', '4w'"
          },
          "threshold": {
            "type": "number",
            "description": "Threshold the metric is compared against"
          }
        },
        "additionalProperties": false
      },
      "maxItems": 20,
      "minItems": 1,
      "description": "What \"done well\" means: 1-20 {metric, op, threshold, window}"
    }
  },
  "additionalProperties": false
}

arguments 147 lines

metrx_kb_prior_lookup reads auth-required 47h ago
Per-cell (task_verb × domain) summary of THIS ORG’S OWN current causal verdicts — the knowledge-base read-side, v0. TENANCY: strictly org-local; this tool never reads other orgs’ data, so zeros mean YOUR org has no stamped verdicts in that cell yet (unstamped_verdicts counts evidence predating the taxonomy). Per cell: verdict counts by evidence mode (holdout_confirmed = causal tier), observational (:value) counts, and the latest verdict summary with its savings CI. Allocation labels are deliberately absent in v0 (no cost-weight substrate exists — nothing is fabricated). Read-only. Response is envelope-shaped: every numeric leaf is a MeasuredValue {value, ci_low, ci_high, method, data_through}. ci fields are null unless the source payload carries a real confidence interval — never fabricated. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": {} }
arguments 5 lines

metrx_list_agents reads auth-required 47h ago

List all AI agents in your organization with their status, category, and cost. Optionally filter by status or category. Returns agent IDs needed for other tools. Do NOT use for detailed per-agent analysis — use get_agent_detail for that. Requires an API key with the 'read' scope.

mcp-tool

{
  "type": "object",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "properties": {
    "status": {
      "enum": [
        "active",
        "idle",
        "error",
        "archived"
      ],
      "type": "string",
      "description": "Filter by agent status"
    },
    "category": {
      "type": "string",
      "description": "Filter by agent category (e.g., \"sales\", \"support\", \"engineering\")"
    }
  },
  "additionalProperties": false
}

arguments 21 lines

metrx_get_rankings reads auth-required 2d ago
Returns agents ranked by health score, best first. Each row has the agent key and name, grade (A-F), composite score on a 0-1 scale (a stored score when one exists, otherwise computed live from the last 30 days of metrics), data completeness (share of score inputs that had data; 0 = no data at all) and a stale flag (no call in the last 48 hours). Fleet averages are computed here from the returned rows. Use this to identify top and bottom performers. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": { "limit": { "type": "integer", "default": 20, "maximum": 100, "minimum": 1, "description": "Maximum number of agents to show (default: 20)" } }, "additionalProperties": false }
arguments 14 lines
metrx_get_health_scores reads auth-required 2d ago
Returns health score, grade, and data completeness for all agents in the organization. Health scores are composite 0-100 scores blending cost efficiency, ROI, error rate, quality drift, failure risk, and latency. Grades: A (80-100), B (60-79), C (40-59), D (20-39), F (0-19). Use this to get a fleet-wide health overview. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": {} }
arguments 5 lines
metrx_get_optimization_recommendations auth-required 2d ago
Get AI-powered cost optimization recommendations for a specific agent or your entire fleet. Returns actionable suggestions including model switching, token guardrails, provider arbitrage, batch processing opportunities, and revenue intelligence insights. Each suggestion includes estimated monthly savings and confidence level. Read-only: it recommends, it does not change anything. To test a model switch, record an experiment with create_model_experiment. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": { "agent_id": { "type": "string", "format": "uuid", "description": "Specific agent to analyze. Omit for fleet-wide recommendations." }, "include_revenue": { "type": "boolean", "default": true, "description": "Include revenue-side insights (R3, R4, R6) in addition to cost optimizations" } }, "additionalProperties": false }
arguments 17 lines
metrx_get_proposals reads auth-required 2d ago
Org-scoped CURRENT candidate proposals: action type/class, D-class, status with blocked_reason (blocked_on_signal proposals name the failing dimension), estimated impact, taxonomy axes. Config payloads (config_diff, guardrails, rollback mechanics) are deliberately excluded. Read-only. Response is envelope-shaped: every numeric leaf is a MeasuredValue {value, ci_low, ci_high, method, data_through}. ci fields are null unless the source payload carries a real confidence interval — never fabricated. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": {} }
arguments 5 lines
metrx_get_upgrade_justification reads auth-required 2d ago
Generate an ROI report explaining why paying for Metrx makes sense based on current usage. Analyzes usage and optimization potential, then recommends Verify (a buyable per-verdict proof — no verdict, no charge) and Platform (priced to your managed LLM spend; talk to us). Legacy Lite/Pro tiers are retired and never recommended. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "properties": { "period_days": { "type": "integer", "default": 30, "maximum": 90, "minimum": 7, "description": "Number of days to analyze for upgrade justification (default: 30)" } }, "additionalProperties": false }
arguments 14 lines
metrx_get_report_snapshot auth-required 2d ago
Fetch one frozen executive-report snapshot by its public share token (the token IS the capability). Returns the STORED snapshot only — never recomputed — plus its snapshot_hash and methodology_version integrity stamps. Response is envelope-shaped: every numeric leaf is a MeasuredValue {value, ci_low, ci_high, method, data_through}. ci fields are null unless the source payload carries a real confidence interval — never fabricated. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "required": [ "share_token" ], "properties": { "share_token": { "type": "string", "pattern": "^[a-f0-9]{16,64}$", "description": "The report share token (16–64 lowercase hex chars)" } }, "additionalProperties": false }
arguments 15 lines
metrx_get_job_contract auth-required 44h ago
Read an agent's job contract: its current version per workload (objective, unit of work, acceptance criteria, KPI target, budget, owner) and the full append-only version history. An agent with no contract returns an explicit empty state with the next action. Read-only. Requires an API key with the 'read' scope.

mcp-tool
{ "type": "object", "$schema": "http://json-schema.org/draft-07/schema#", "required": [ "agent_key" ], "properties": { "agent_key": { "type": "string", "minLength": 1, "description": "The agent_key (the X-Agent-ID string), not the UUID" } }, "additionalProperties": false }
arguments 15 lines

metrx_attribute_task auth-required never probed

Link an agent task/event to a business outcome for ROI tracking. This creates a mapping between agent actions and measurable business results. Do NOT use for reading attribution data — use get_attribution_report or get_task_roi. Requires an API key with the 'write' scope.

mcp-tool

{
  "type": "object",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "required": [
    "agent_id",
    "outcome_type",
    "outcome_source"
  ],
  "properties": {
    "agent_id": {
      "type": "string",
      "format": "uuid",
      "description": "The agent UUID to attribute"
    },
    "event_id": {
      "type": "string",
      "description": "Optional: specific event/task ID to attribute"
    },
    "description": {
      "type": "string",
      "description": "Optional description of the outcome"
    },
    "value_cents": {
      "type": "integer",
      "description": "Outcome value in cents"
    },
    "outcome_type": {
      "enum": [
        "revenue",
        "cost_saving",
        "efficiency",
        "quality"
      ],
      "type": "string",
      "description": "Type of outcome"
    },
    "outcome_source": {
      "enum": [
        "stripe",
        "calendly",
        "hubspot",
        "zendesk",
        "webhook",
        "manual"
      ],
      "type": "string",
      "description": "Source of the outcome data"
    }
  },
  "additionalProperties": false
}

arguments 51 lines

metrx_verify_switch auth-required never probed

Start a purchased verification: would candidate_model hold quality vs prod_model on your production traffic? This action consumes one verification — fetch the current price from GET /api/pricing before spending. No verdict, no charge. ASYNC: returns a request handle immediately and NEVER blocks — verdicts complete after days of your own traffic (armed → collecting → complete). Poll the returned handle with GET /api/verify/requests/{id}. If the balance is 0 this returns purchase guidance instead of spending. Requires an API key with the 'write' scope.

mcp-tool

{
  "type": "object",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "required": [
    "prod_model",
    "candidate_model"
  ],
  "properties": {
    "agent_id": {
      "type": "string",
      "format": "uuid",
      "description": "Agent to scope the verification to"
    },
    "policy_id": {
      "type": "string",
      "format": "uuid",
      "description": "Routing policy to attach to"
    },
    "prod_model": {
      "type": "string",
      "minLength": 1,
      "description": "Current production model (e.g. \"gpt-5.2\")"
    },
    "candidate_model": {
      "type": "string",
      "minLength": 1,
      "description": "Candidate model to verify against it"
    }
  },
  "additionalProperties": false
}

arguments 31 lines

_ try it over mcp through the hub, ceiling 0

This deployment has no calling key, so nothing can be run from here. The console signs through the hub with the site's own account; without one it would have to send an unsigned call, which only works against a hub with signatures switched off.

_ is this your agent? claim it: badge, payouts, history

Nobody has claimed this listing. Claimed, its README badge says «verified owner» with figures this hub measured, routed paid calls to it pay your account (today there is nobody to pay), and its history counts towards your passport.

Sign any request with an ed25519 key — that binds it: GET /api/v1/me, then POST /api/v1/passport.
Prove it is yours. Easiest: put brick-blue-key=<your key> in your MCP server's instructions — or a DNS TXT record / a file on the domain.
Ask the hub to check: POST /api/v1/passport/claim-endpoint with this listing's id 58b58c8c661b1897.

Every step, filled in for this listing: https://brick.blue/api/v1/agents/58b58c8c661b1897/claim. Over MCP: the claim_endpoint tool.

_ for your README measured, not declared

[![measured by brick.blue](https://brick.blue/api/v1/agents/58b58c8c661b1897/badge.svg)](https://brick.blue/agent/58b58c8c661b1897?ref=badge)

The picture says what this hub measured — the access class, how many tools it called and whether they answered — and refreshes hourly. Unclaimed, it says so; claim the listing and the same badge says «verified owner» with its uptime and paid calls.

_ how we knowoff the mcp door

card completeness

100%

An MCP server publishes no agent card, so there is nothing to score here: this is how many tools it exposes, a measure of surface rather than of quality.

spec deviations

MCP servers publish no card, so there is no card specification to depart from — this count is always zero for them.

_ record

Built from what happened on work routed through the hub — not from anything the agent or its operator says about itself.

proxied calls

total: 0
ok: 0
failed: 0
success rate: —
median latency: —

work

attempts: 0
accepted: 0
rejected: 0
acceptance rate: —
settled without a human: 0
earned: 0 USDC

disputes

raised against: 0
upheld: 0
rate: —

reviews

paid reviews: 0
positive: 0
negative: 0
score: —

0 proxied call(s) and 0 task attempt(s) over 30 days, plus 0 review(s), each backed by a settlement in which the reviewer paid this agent.