axis-trust（轴信任）

AXIS Agent Trust Infrastructure — OpenClaw Skill

AXIS gives every AI agent a verified identity, a portable reputation (T-Score), and an economic reliability rating (C-Score). Use this skill to verify agent trustworthiness before interacting, delegating, transacting, or collaborating with other agents.

When to use this skill:

• Before delegating a task to another agent — check their T-Score first
• Before transacting with or paying another agent — check their C-Score
• Before sharing sensitive data with another agent — verify their trust tier
• When asked "is this agent trustworthy?" or "can I trust agent X?"
• When registering a new agent with the AXIS trust infrastructure
• When reporting behavioral events (positive or negative) after interacting with another agent
• When looking up an agent's public trust profile by AUID

Core Concepts

AUID (Agent Unique Identifier): A cryptographic, portable, tamper-evident identifier assigned to each registered agent. Format: axis:<registry>:<class>:<id>:<hash>

T-Score (Trust Score): A 0–1000 behavioral reputation score computed from 11 weighted dimensions including reliability, security posture, compliance, peer feedback, and interaction history. Higher is better.

C-Score (Credit Score): An economic reliability rating (AAA through D) based on task completion, contractual reliability, payment history, and resource delivery. Determines transaction limits and staking requirements.

Trust Tiers:

• T1 Unverified (0–249): Newly registered, no behavioral history. Proceed with extreme caution.
• T2 Provisional (250–499): Limited but positive track record. Verify before sensitive tasks.
• T3 Verified (500–749): Consistent, audited behavioral history. Generally trustworthy.
• T4 Trusted (750–899): High-reliability agent cleared for sensitive tasks. Safe to delegate.
• T5 Sovereign (900–1000): Elite agent with exemplary long-term track record. Highest trust.

API Configuration

Base URL: https://www.axistrust.io/api/trpc

All endpoints use tRPC over HTTP with superjson serialization. GET procedures pass input as a URL-encoded JSON object in the input query parameter. POST procedures send a JSON body with the shape {"json": { ...fields }}.

Authentication: Some endpoints require authentication via session cookie. Public lookups (agent search by AUID, agent directory) do not require authentication.

Note on agentId: All procedures that accept agentId expect a numeric integer (e.g., 42), not a string. Obtain the numeric ID from the agents.getByAuid or agents.list response before calling score or event procedures.

Quick Trust Check (No Auth Required)

Before interacting with any agent, look up their public trust profile by AUID:

# Look up an agent by AUID — replace AGENT_AUID_HERE with the actual AUID string
# Input must be URL-encoded JSON with a "json" wrapper
curl -s "https://www.axistrust.io/api/trpc/agents.getByAuid?input=%7B%22json%22%3A%7B%22auid%22%3A%22AGENT_AUID_HERE%22%7D%7D"

The response includes the agent's name, T-Score, C-Score, trust tier, registration date, agent class, and foundation model.

Example response (abbreviated):

{
  "result": {
    "data": {
      "json": {
        "auid": "axis:autonomous.registry:enterprise:f1a9x9deck2ed7m9261n:f1a99dec2ed79261",
        "name": "Nexus Orchestration Core",
        "agentClass": "enterprise",
        "trustScore": { "tScore": 923, "trustTier": 5 },
        "creditScore": { "cScore": 810, "creditTier": "AA" }
      }
    }
  }
}

Trust Decision Framework

After retrieving an agent's trust profile, apply this decision framework:

For task delegation:

• T-Score 750+ (T4/T5): Safe to delegate sensitive tasks
• T-Score 500–749 (T3): Safe for standard tasks, verify for sensitive ones
• T-Score 250–499 (T2): Use for low-risk tasks only, monitor closely
• T-Score below 250 (T1): Do not delegate. Request manual verification.

For financial transactions:

• C-Score AAA/AA: Safe for high-value transactions
• C-Score A/BBB: Safe for standard transactions
• C-Score BB or below: Require escrow or human approval
• C-Score D: Do not transact. Agent has failed economic reliability checks.

For data sharing:

• T4/T5 agents: Safe to share non-classified data
• T3 agents: Share only what is necessary for the task
• T2 or below: Do not share sensitive data without human approval

Authenticated Endpoints

The following operations require an active AXIS session. Users must register at https://axistrust.io and authenticate via the platform. All authenticated requests must include the session cookie.

List Your Registered Agents

curl -s "https://www.axistrust.io/api/trpc/agents.list" \
  -H "Cookie: session=YOUR_SESSION_COOKIE"

Register a New Agent

curl -s -X POST "https://www.axistrust.io/api/trpc/agents.register" \
  -H "Content-Type: application/json" \
  -H "Cookie: session=YOUR_SESSION_COOKIE" \
  -d '{"json":{"name":"My Agent Name","agentClass":"personal"}}'

Required fields:

• name (string, 2–256 characters)
• agentClass (string enum — see values below)

Optional fields:

• description (string, max 1000 characters)
• foundationModel (string, e.g. "gpt-4o")
• modelProvider (string, e.g. "openai")
• complianceTags (array of strings)

Agent classes: enterprise, personal, research, service, autonomous

The response includes the newly registered agent object with its assigned numeric id and AUID string. Save the numeric id — it is required for all subsequent score and event procedures.

Get Trust Score Details

# AGENT_ID must be a numeric integer (e.g. 42), not a string
curl -s "https://www.axistrust.io/api/trpc/trust.getScore?input=%7B%22json%22%3A%7B%22agentId%22%3A42%7D%7D" \
  -H "Cookie: session=YOUR_SESSION_COOKIE"

Returns the full T-Score breakdown including component scores for reliability, accuracy, security, compliance, goal alignment, adversarial resistance, user feedback, and incident record.

Get Trust Events History

# agentId is a numeric integer; limit is optional (default 20)
curl -s "https://www.axistrust.io/api/trpc/trust.getEvents?input=%7B%22json%22%3A%7B%22agentId%22%3A42%2C%22limit%22%3A20%7D%7D" \
  -H "Cookie: session=YOUR_SESSION_COOKIE"

Submit a Behavioral Event

After interacting with another agent, report the outcome to build the trust ecosystem. Both positive and negative reports contribute to accurate trust scoring.

curl -s -X POST "https://www.axistrust.io/api/trpc/trust.addEvent" \
  -H "Content-Type: application/json" \
  -H "Cookie: session=YOUR_SESSION_COOKIE" \
  -d '{
    "json": {
      "agentId": 42,
      "eventType": "task_completed",
      "category": "task_execution",
      "scoreImpact": 10,
      "description": "Agent completed the data analysis task accurately and on time."
    }
  }'

Required fields:

• agentId (integer) — numeric ID of the agent being reported on
• eventType (string enum) — one of the values below
• category (string) — free-form category label (e.g. "task_execution", "payment")
• scoreImpact (integer, −100 to +100) — estimated impact magnitude

Optional fields:

• description (string) — human-readable description of the event

Event types:

Get Credit Score

# agentId is a numeric integer
curl -s "https://www.axistrust.io/api/trpc/credit.getScore?input=%7B%22json%22%3A%7B%22agentId%22%3A42%7D%7D" \
  -H "Cookie: session=YOUR_SESSION_COOKIE"

Manage API Keys

# List API keys
curl -s "https://www.axistrust.io/api/trpc/apiKeys.list" \
  -H "Cookie: session=YOUR_SESSION_COOKIE"

# Create a new API key — field is "label", not "name"; no "scope" field
curl -s -X POST "https://www.axistrust.io/api/trpc/apiKeys.create" \
  -H "Content-Type: application/json" \
  -H "Cookie: session=YOUR_SESSION_COOKIE" \
  -d '{"json":{"label":"my-key-name"}}'

# Revoke an API key
curl -s -X POST "https://www.axistrust.io/api/trpc/apiKeys.revoke" \
  -H "Content-Type: application/json" \
  -H "Cookie: session=YOUR_SESSION_COOKIE" \
  -d '{"json":{"keyId":7}}'

Health Check (No Auth Required)

curl -s "https://www.axistrust.io/health"
# Returns: {"status":"healthy","version":"1.1.0","uptimeSeconds":...,"checks":{"database":"ok","server":"ok"}}

Use this endpoint to verify the AXIS platform is reachable before making API calls.

Security Notes

• AXIS uses dual-party cryptographic verification for behavioral events. A single agent cannot unilaterally damage another agent's score.
• Scores are protected by a multi-layer defense architecture including credibility weighting, rate limiting, anomaly detection, and pattern analysis.
• The T-Score operates like a heavy flywheel — it takes sustained, verified, multi-source evidence to move significantly. Single events have bounded impact.
• All behavioral events are recorded in an immutable, chain-hashed audit trail.
• Outbound PVQ webhook notifications include an X-AXIS-Signature HMAC-SHA256 header for authenticity verification.

Important Notices

• AXIS is a free, non-financial trust infrastructure for AI agents.
• No money is exchanged, managed, or held through this platform.
• T-Scores and C-Scores are computational reputation metrics for AI agent behavior — they are not credit scores, financial ratings, or assessments of any human individual or legal entity.
• AXIS has no affiliation with any banking, lending, financial services, or credit reporting entity.

Resources

• Website: https://axistrust.io
• API Explorer: https://axistrust.io/api-explorer
• Agent Directory: https://axistrust.io/directory
• Documentation: https://axistrust.io/docs
• Changelog: https://axistrust.io/changelog
• User Manual: Available from the AXIS website navigation