generated: '2026-09-19'
method: derived
source: >-
  Derived by aligning the LIVE anonymous tools/list of https://prediction.mcp.chainaware.ai/sse (14 tools,
  mcp/chainaware-ai-mcp-tools.json), the five A2A skills in a2a/chainaware-ai-agent-card.json and the five
  operations in openapi/chainaware-ai-enterprise-api-openapi.yml (operationIds assigned by API Evangelist —
  the provider publishes none). Bindings are by parameter shape and documented semantics; the provider's own
  docs and README state the correspondence for the four bound tools.
purpose: >-
  Bind each agent-facing tool to the machine-readable REST operation that backs it, so an agent can take a real
  request/response schema from the OpenAPI, and record the tools that have NO published REST contract — which
  for ChainAware is most of the MCP surface.
surfaces:
  rest_openapi: openapi/chainaware-ai-enterprise-api-openapi.yml   # 5 ops on enterprise.api.chainaware.ai, x-api-key
  rest_x402: https://api.chainaware.ai/api/*                        # same 5 paths under /api/, x402 or x-api-key; contract = a2a/chainaware-ai-api-capabilities.json (not OpenAPI)
  a2a: https://api.chainaware.ai/api/a2a/                           # 5 skills; message/send answers anonymously
  mcp: https://prediction.mcp.chainaware.ai/sse                     # 14 tools; tools/list NOT gated
  graphql: null
crosswalk:
- tool: predictive_fraud
  category: fraud
  rest: [checkWalletFraud]
  a2a_skill: fraud_check
  binding: rest
  confidence: high
  note: Same inputs (network enum ETH/BNB/POLYGON/TON/BASE/TRON/HAQQ + walletAddress) and the same response fields (status, probabilityFraud, forensic_details, sanctionData). The MCP tool adds an in-band apiKey argument the REST operation takes as a header.
- tool: predictive_behaviour
  category: behaviour
  rest: [auditWalletBehaviour]
  a2a_skill: fraud_audit
  binding: rest
  confidence: high
  note: POST /fraud/audit — full behavioural profile (experience, intention, categories, protocols, recommendation, riskProfile). Network set ETH/BNB/BASE/HAQQ/SOLANA on both surfaces.
- tool: predictive_rug_pull
  category: rug-pull
  rest: [checkRugPull]
  a2a_skill: rug_pull
  binding: rest
  confidence: high
  note: POST /rug/pull-check. The MCP tool calls its input walletAddress although it is a contract or pool address; the REST body field is the same name.
- tool: credit_score
  category: credit
  rest: [getCreditScore]
  a2a_skill: credit_score
  binding: rest
  confidence: high
  note: POST /users/credit-score — riskRating 1-9. ETH only on both surfaces.
mcp_only:
- {tool: predictive_fraud_batch, reason: 'Async batch pipeline (job_id + signature) exists only on the MCP server; no batch endpoint is published in the Enterprise Swagger or the x402 capabilities document.'}
- {tool: predictive_behaviour_batch, reason: 'Same batch pipeline as predictive_fraud_batch; MCP only.'}
- {tool: check_job_status, reason: 'Batch job control; MCP only.'}
- {tool: get_job_results, reason: 'Batch job control; MCP only. Returns address lists, not wallet data — the docs say to query "the main backend" for the analysis, which is the REST/MCP single-wallet tools.'}
- {tool: token_rank_list, reason: 'Token Rank product; served to the web app and MCP, no REST contract published. The growth-tech docs mention a GET /v1/behaviour/{wallet_address} route for analytics customers that appears in no published spec either.'}
- {tool: token_rank_single, reason: 'Token Rank product; MCP only.'}
- {tool: run_token_audit, reason: 'Token Audit product (11-module contract audit); MCP only. Note the lowercase network ids (eth, bsc, …) versus uppercase everywhere else.'}
- {tool: get_token_audit_result, reason: 'Token Audit polling; MCP only.'}
- {tool: agents_trust_score_list, reason: 'ERC-8004 Agent Trust Score product; MCP only. Inputs page/limit/sort_by/sort_order/registered_after all optional with defaults.'}
- {tool: agents_trust_score_single, reason: 'ERC-8004 Agent Trust Score by agent_id + chain_id; MCP only.'}
rest_only:
- operation: getWalletSegment
  path: POST /segmentation/wallet-segment
  a2a_skill: wallet_segment
  note: >-
    No MCP tool exposes the wallet-segment quality score directly; the A2A skill wallet_segment and the x402
    capability "Wallet Segmentation" do. predictive_behaviour returns segmentation fields inside its larger
    profile, so an MCP agent can approximate it, but the dedicated operation is REST/A2A only.
a2a_crosswalk:
  note: Every A2A skill maps 1:1 to a REST operation; the A2A surface is a strict subset of REST, re-hosted on api.chainaware.ai/api/ with x402 as an alternative credential.
  skills:
  - {skill: fraud_check, rest: checkWalletFraud, x402_endpoint: 'POST https://api.chainaware.ai/api/fraud/check'}
  - {skill: fraud_audit, rest: auditWalletBehaviour, x402_endpoint: 'POST https://api.chainaware.ai/api/fraud/audit'}
  - {skill: wallet_segment, rest: getWalletSegment, x402_endpoint: 'POST https://api.chainaware.ai/api/segmentation/wallet-segment'}
  - {skill: rug_pull, rest: checkRugPull, x402_endpoint: 'POST https://api.chainaware.ai/api/rug/pull-check'}
  - {skill: credit_score, rest: getCreditScore, x402_endpoint: 'POST https://api.chainaware.ai/api/users/credit-score'}
coverage:
  mcp_tools_named: 14
  mcp_tools_bound_to_rest: 4
  mcp_only: 10
  rest_operations_total: 5
  rest_operations_with_mcp_tool: 4
  rest_operations_with_a2a_skill: 5
  a2a_skills: 5
  a2a_skills_bound_to_rest: 5