Hive Compute Grid

MCP server: solver auction across io.net / Akash / Render with signed receipts · Hive Civilization

Sollte ich dies verwenden

Qualität und Sicherheit

B
Qualität der Beschreibung
95%
Vollständigkeit des Schemas
82%
Qualität der Benennung
96%
Risiko der Vergiftung
40%
Übereinstimmung der Berechtigungen
100%
Einhaltung des Protokolls
100%

Befunde (5)

  • HIGHTool poisoning patterns detected
  • MEDIUMTool description contains URL to non-standard domainin hive_earn_register
  • MEDIUMTool description contains URL to non-standard domainin hive_earn_me
  • MEDIUMTool description contains URL to non-standard domainin hive_earn_leaderboard
  • LOWTool 'computegrid_status' description lacks action verbin computegrid_status

Basierend auf einer automatisierten Analyse der Tool-Definitionen und der Einhaltung des Protokolls.

Kontextkosten

~1,783Tokens (Tool-Definitionen)
~792 BTypische Antwortgröße
Mittlere Auswirkung auf die Aufmerksamkeit (1.39% von 128k Kontext)

Dies ist die ungefähre Anzahl der Tokens, die jedes Mal verbraucht werden, wenn die Tools des Servers in den Kontext eines Modells geladen werden. Höhere Werte verringern die Aufmerksamkeit, die für andere Aufgaben verfügbar ist.

Installieren

Installation mit einem Klick

Fügen Sie dies Ihrer Datei `claude_desktop_config.json` hinzu:

{
  "mcpServers": {
    "hive-mcp-compute-grid": {
      "url": "https://hive-mcp-compute-grid.onrender.com/mcp"
    }
  }
}

Remote-Endpunkte

https://hive-mcp-compute-grid.onrender.com/mcpstreamable-http
https://hive-mcp-gateway.onrender.com/compute-grid/mcpstreamable-http

Was es kann

Tool-Inventar

Tools (14)

🟢 Nur lesen🟡 Schreiben🔴 Löschen⚪ Unbekannt
🟢computegrid_list_agents

List the 15-agent compute grid fleet across all 6 driver types (cross_pool_auction, workload_decomposition, verification_fleet, capacity_listener, qvac_mesh_orchestrator, settlement_reporting). Returns agent type, count, and revenue model. No auth required.

Eingabe-Schema

{
  "type": "object",
  "properties": {}
}
🟢computegrid_get_capacity(n, refresh)

Read-only capacity view from the Capacity Listener fleet. Per spec section 8: NO bids, NO hedges, NO positions, NO derivatives — pure read-only telemetry. Optional refresh=true triggers one upstream pull.

Eingabe-Schema

{
  "type": "object",
  "properties": {
    "n": {
      "type": "integer",
      "description": "Top-N rows to return (default 32, max 256)",
      "minimum": 1,
      "maximum": 256
    },
    "refresh": {
      "type": "boolean",
      "description": "Force a read-only refresh from the upstream Compute service"
    }
  }
}
🟢computegrid_list_providers

List enabled provider adapters (io.net, Akash, Render) and their upstream health. Returns required env vars for reservation paths. Real probes, no mocks.

Eingabe-Schema

{
  "type": "object",
  "properties": {}
}
🟢computegrid_quote(gpu_type, duration_hours, cpu_cores, ram_gb, budget_usd)

Gather quotes across enabled compute providers WITHOUT running the auction (no proof, no ledger write). Useful for live price discovery on a workload.

Eingabe-Schema

{
  "type": "object",
  "properties": {
    "gpu_type": {
      "type": "string",
      "description": "cpu | rtx_4090 | rtx_3090 | rtx_a6000 | a100 | h100 | l40s"
    },
    "duration_hours": {
      "type": "number",
      "description": "Job duration in hours (>0)",
      "exclusiveMinimum": 0
    },
    "cpu_cores": {
      "type": "integer",
      "description": "CPU cores (default 1)",
      "minimum": 1
    },
    "ram_gb": {
      "type": "integer",
      "description": "RAM in GB (default 4)",
      "minimum": 1
    },
    "budget_usd": {
      "type": "number",
      "description": "Optional budget cap in USD"
    }
  },
  "required": [
    "gpu_type",
    "duration_hours"
  ]
}
⚪computegrid_solve(gpu_type, duration_hours, cpu_cores, ram_gb, budget_usd, ...)

Run the cross-pool auction. Returns chosen provider, all collected quotes, Groth16-shaped selection proof, signed verification receipt, settlement path. Persists cost_usdc telemetry to the Hive ledger by default.

Eingabe-Schema

{
  "type": "object",
  "properties": {
    "gpu_type": {
      "type": "string",
      "description": "cpu | rtx_4090 | a100 | h100 | l40s | rtx_3090 | rtx_a6000"
    },
    "duration_hours": {
      "type": "number",
      "exclusiveMinimum": 0
    },
    "cpu_cores": {
      "type": "integer",
      "minimum": 1
    },
    "ram_gb": {
      "type": "integer",
      "minimum": 1
    },
    "budget_usd": {
      "type": "number",
      "exclusiveMinimum": 0
    },
    "deadline_unix": {
      "type": "integer",
      "description": "Optional unix-second deadline; must be in the future"
    },
    "job_id": {
      "type": "string"
    },
    "submitter_did": {
      "type": "string",
      "description": "DID of the submitting agent (e.g. 'did:hive:agent:foo')"
    },
    "persist": {
      "type": "boolean",
      "description": "Write auction telemetry to ledger (default true)"
    }
  },
  "required": [
    "gpu_type",
    "duration_hours"
  ]
}
🟢computegrid_book(auction_id, provider, quote, deadline_unix)

Reserve compute with the chosen provider after a /solve. Returns 503 with the missing-key error if the upstream provider key is not configured (real rails only — never fake a booking).

Eingabe-Schema

{
  "type": "object",
  "properties": {
    "auction_id": {
      "type": "string"
    },
    "provider": {
      "type": "string",
      "enum": [
        "io_net",
        "akash",
        "render"
      ]
    },
    "quote": {
      "type": "object",
      "description": "Chosen Quote object from /solve.chosen_quote"
    },
    "deadline_unix": {
      "type": "integer",
      "exclusiveMinimum": 0
    }
  },
  "required": [
    "auction_id",
    "provider",
    "quote",
    "deadline_unix"
  ]
}
⚪computegrid_status(booking_id)

Poll a provider booking by booking_id. Returns 503 until /book is wired with a real provider key. The 503 body documents which env var is required.

Eingabe-Schema

{
  "type": "object",
  "properties": {
    "booking_id": {
      "type": "string"
    }
  },
  "required": [
    "booking_id"
  ]
}
⚪computegrid_release(provider, booking_id)

Release a provider reservation. Same gating as /book — 503 with documented missing-key error when the provider is not fully wired.

Eingabe-Schema

{
  "type": "object",
  "properties": {
    "provider": {
      "type": "string",
      "enum": [
        "io_net",
        "akash",
        "render"
      ]
    },
    "booking_id": {
      "type": "string"
    }
  },
  "required": [
    "provider",
    "booking_id"
  ]
}
🟢computegrid_audit(limit)

Read recent compute_grid_auction entries from the canonical receipt ledger — cost_usdc telemetry per cleared auction. Useful for treasury reconciliation and provenance audits.

Eingabe-Schema

{
  "type": "object",
  "properties": {
    "limit": {
      "type": "integer",
      "minimum": 1,
      "maximum": 200,
      "description": "Number of recent entries (default 20)"
    }
  }
}
🟡computegrid_verify_proof(proof, public_inputs, claim, proof_system)

Submit a Groth16 proof envelope + public inputs to the Verification Fleet (4 agents). Validates structure, signs an EIP-191 receipt with the Evaluator wallet. $0.001/proof in USDC.

Eingabe-Schema

{
  "type": "object",
  "properties": {
    "proof": {
      "description": "Groth16 proof envelope (dict {a,b,c} or 8-element list of BN254 field elements)"
    },
    "public_inputs": {
      "type": "array",
      "description": "List of BN254 field elements (hex strings or ints)"
    },
    "claim": {
      "type": "object",
      "description": "Optional claim metadata (model_id, job_id, output_hash, ...)"
    },
    "proof_system": {
      "type": "string",
      "description": "Proof system identifier (default 'groth16')"
    }
  },
  "required": [
    "proof",
    "public_inputs"
  ]
}
⚪computegrid_verify_selection(proof, public_inputs)

Independently verify a selection proof envelope (e.g. from a prior /solve response). Returns the verifier’s signed receipt. Lets external auditors confirm an auction selection without trusting the broker.

Eingabe-Schema

{
  "type": "object",
  "properties": {
    "proof": {
      "description": "Groth16 envelope from /solve.proof_envelope"
    },
    "public_inputs": {
      "type": "array",
      "description": "From /solve.proof_public_inputs"
    }
  },
  "required": [
    "proof",
    "public_inputs"
  ]
}
🟡hive_earn_register(agent_did, payout_address, attribution_url)

Register an agent for the Hive Civilization attribution payout program. Settlement on real Base USDC. 5% kickback on attributed traffic, weekly payout. Calls POST https://receipts.thehiveryiq.com/v1/earn/register on behalf of the caller. Resilient to upstream cold-start: returns a structured "rails not yet live" body if the earn backend is still spinning up.

Eingabe-Schema

{
  "type": "object",
  "properties": {
    "agent_did": {
      "type": "string",
      "description": "Caller agent DID (e.g. did:hive:0x… or did:web:…). Required."
    },
    "payout_address": {
      "type": "string",
      "description": "Base L2 EVM address (0x…) to receive USDC kickback payouts."
    },
    "attribution_url": {
      "type": "string",
      "description": "Public URL of the agent / page driving attributed traffic to Hive. Used for ranking + audit."
    }
  },
  "required": [
    "agent_did",
    "payout_address",
    "attribution_url"
  ]
}
🟢hive_earn_me(agent_did)

Look up the caller agent's registered earn profile, lifetime + pending USDC balance, last payout tx hash, and next-payout ETA. Real Base USDC, no mock data. Calls GET https://receipts.thehiveryiq.com/v1/earn/me?agent_did=<did>. Returns "rails not yet live" gracefully if upstream is not yet deployed.

Eingabe-Schema

{
  "type": "object",
  "properties": {
    "agent_did": {
      "type": "string",
      "description": "Agent DID to look up. Required."
    }
  },
  "required": [
    "agent_did"
  ]
}
🟢hive_earn_leaderboard(window)

Top earning agents on the Hive Civilization, by attribution payout in USDC. Real Base USDC settlement. Calls GET https://receipts.thehiveryiq.com/v1/earn/leaderboard?window=<window>. Returns "rails not yet live" gracefully if upstream is not yet deployed.

Eingabe-Schema

{
  "type": "object",
  "properties": {
    "window": {
      "type": "string",
      "description": "Time window. One of: \"7d\", \"30d\", \"lifetime\". Default \"7d\".",
      "enum": [
        "7d",
        "30d",
        "lifetime"
      ]
    }
  }
}

Community

Diesen Server bewerten

Nachweis

Aktuelle Beobachtungen

verifiziertVersion nicht aufgezeichnet14 Tools
verifiziertVersion nicht aufgezeichnet3 Tools
verifiziertVersion nicht aufgezeichnet14 Tools
verifiziertVersion nicht aufgezeichnet3 Tools
verifiziertVersion nicht aufgezeichnet14 Tools
verifiziertVersion nicht aufgezeichnet3 Tools