Best Practices (ERC-8004)

Trust isn't a feature — it's infrastructure. ERC-8004 defines identity, reputation, and validation primitives for autonomous agents, but trustworthy systems depend on how you publish metadata, record feedback, and defend against abuse.

This page is implementation-agnostic. We separate ERC-8004 Standard (interoperability requirements) from DenLabs Profile (our recommended defaults).

1.Establish a Clear Agent Identity

Goal: Make every actor discoverable, stable, and attributable.

ERC-8004 Standard

•Register a stable AgentID (do not rotate identifiers to reset history).
•Publish an AgentURI that is reachable and durable.
•Ensure metadata is accurate and versioned.

DenLabs Profile

•One canonical identity per agent + one public 'Agent Card' page.
•AgentURI should include: name, description, version, capabilities[], endpoints (HTTP/MCP/A2A) with versions, registrations[] (chain + contract addresses), optional attestations[] (e.g., Self.xyz verification).

Anti-patterns

Creating a new AgentID per release.
Metadata that cannot be resolved reliably (broken URLs, ephemeral hosts).

2.Design Honest & Verifiable Reputation

Goal: Reputation must reflect real interactions with evidence and context.

ERC-8004 Standard

•Keep minimal signals on-chain (IDs, pointers, hashes).
•Store rich feedback off-chain via URI + integrity hash.

DenLabs Profile

•Every feedback record SHOULD include: timestamp (ISO 8601), context (event/session/task identifiers), actor identifier (wallet/session/proof reference), structured outcome (labels, score, tags), evidence[] (tx hashes, links, attachments, logs).
•Use consistent schemas so third parties can index and compare outcomes.

Anti-patterns

Star ratings with no context.
Reputation based only on unverifiable self-reports.

3.Use Validation for High-Impact Work

Goal: Add measurable confidence for sensitive or high-value operations.

When to require validation

•High-value payouts / grant decisions
•Governance voting eligibility
•Security-sensitive workflows
•Irreversible actions

DenLabs Profile

•A validation record SHOULD include: target (what was validated), validator identity (who validated), methodology + criteria, result + confidence score, reproducible artifacts (report hashes, logs, signed statements).

Anti-patterns

'Trust me' validations without methodology or evidence.

4.Optimize On-Chain vs Off-Chain Data

Goal: Scale without losing integrity.

Recommended split

•On-chain: AgentID, pointers (URI), hashes, minimal counters/flags
•Off-chain: heavy payloads (feedback bodies, reports, attachments, analytics)

Why it matters

•Cost and throughput optimization
•Indexability for third-party consumers
•Privacy controls where needed
•Upgradeability without migrations

5.Publish Clear Interfaces & Documentation

Goal: Interoperability is a trust multiplier.

Required

•Versioned endpoints (/v1/...)
•Clear capability descriptions (inputs/outputs, constraints)
•Error taxonomy + rate limits

Provide

•'Happy-path' examples
•Explicit assumptions (what you do NOT protect against)
•Stability policy (how breaking changes are handled)

6.Long-Term Trust & Abuse Resistance

Goal: Trust compounds; abuse scales.

Recommended defenses

•Rate limiting: cap submissions per session and per wallet
•Sybil resistance: require stronger proofs for higher trust tiers (e.g., Self.xyz)
•Evidence gating: higher scores require stronger evidence
•Recency weighting: recent behavior matters more than old history
•Anomaly detection: bursts, repeated patterns, recycled content

Anti-patterns

New identities reaching max trust instantly.
Unlimited submissions per session.

Implementation Checklist

Templates

AgentURI (minimal useful example)

{
  "name": "DenLabs Trust Scorer",
  "description": "Trust scoring for event feedback operations.",
  "version": "1.0.0",
  "capabilities": ["trust_score", "feedback_filtering"],
  "endpoints": [
    { "type": "http", "url": "https://denlabs.vercel.app/api/scan/8004", "version": "v1" }
  ],
  "registrations": [
    { "chain": "eip155:1", "contract": "0x...", "standard": "erc-8004" }
  ]
}

Feedback record (off-chain)

{
  "subject": { "type": "agent", "id": "AgentID:123" },
  "actor": { "type": "wallet", "id": "eip155:1:0xabc..." },
  "timestamp": "yyyy-mm-ddThh:mm:ssZ",
  "context": { "event": "DenLabs Demo Day", "session": "retro-pack-01" },
  "rating": { "label": "useful", "value": 1 },
  "evidence": [
    { "type": "tx", "ref": "0xdeadbeef..." },
    { "type": "url", "ref": "https://..." }
  ],
  "notes": "Actionable feedback with reproducible details."
}

Validation record (off-chain)

{
  "target": { "type": "feedback_item", "id": "feedback:0xhash" },
  "validator": { "type": "entity", "id": "validator:denlabs" },
  "timestamp": "yyyy-mm-ddThh:mm:ssZ",
  "method": { "name": "manual_review", "criteria": ["reproducible", "non-spam", "contextual"] },
  "result": { "status": "pass", "confidence": 0.86 },
  "artifacts": [{ "type": "hash", "ref": "sha256:..." }]
}

Library > Trust Scoring (8004)

Trust Scoring (8004)

Trust verification for Event Feedback Ops.

Best Practices (ERC-8004)

This page is implementation-agnostic. We separate ERC-8004 Standard (interoperability requirements) from DenLabs Profile (our recommended defaults).

1.Establish a Clear Agent Identity

Goal: Make every actor discoverable, stable, and attributable.

ERC-8004 Standard

•Register a stable AgentID (do not rotate identifiers to reset history).
•Publish an AgentURI that is reachable and durable.
•Ensure metadata is accurate and versioned.

DenLabs Profile

•One canonical identity per agent + one public 'Agent Card' page.
•AgentURI should include: name, description, version, capabilities[], endpoints (HTTP/MCP/A2A) with versions, registrations[] (chain + contract addresses), optional attestations[] (e.g., Self.xyz verification).

Anti-patterns

Creating a new AgentID per release.
Metadata that cannot be resolved reliably (broken URLs, ephemeral hosts).

2.Design Honest & Verifiable Reputation

Goal: Reputation must reflect real interactions with evidence and context.

ERC-8004 Standard

•Keep minimal signals on-chain (IDs, pointers, hashes).
•Store rich feedback off-chain via URI + integrity hash.

DenLabs Profile

•Every feedback record SHOULD include: timestamp (ISO 8601), context (event/session/task identifiers), actor identifier (wallet/session/proof reference), structured outcome (labels, score, tags), evidence[] (tx hashes, links, attachments, logs).
•Use consistent schemas so third parties can index and compare outcomes.

Anti-patterns

Star ratings with no context.
Reputation based only on unverifiable self-reports.

3.Use Validation for High-Impact Work

Goal: Add measurable confidence for sensitive or high-value operations.

When to require validation

•High-value payouts / grant decisions
•Governance voting eligibility
•Security-sensitive workflows
•Irreversible actions

DenLabs Profile

•A validation record SHOULD include: target (what was validated), validator identity (who validated), methodology + criteria, result + confidence score, reproducible artifacts (report hashes, logs, signed statements).

Anti-patterns

'Trust me' validations without methodology or evidence.

4.Optimize On-Chain vs Off-Chain Data

Goal: Scale without losing integrity.

Recommended split

•On-chain: AgentID, pointers (URI), hashes, minimal counters/flags
•Off-chain: heavy payloads (feedback bodies, reports, attachments, analytics)

Why it matters

•Cost and throughput optimization
•Indexability for third-party consumers
•Privacy controls where needed
•Upgradeability without migrations

5.Publish Clear Interfaces & Documentation

Goal: Interoperability is a trust multiplier.

Required

•Versioned endpoints (/v1/...)
•Clear capability descriptions (inputs/outputs, constraints)
•Error taxonomy + rate limits

Provide

•'Happy-path' examples
•Explicit assumptions (what you do NOT protect against)
•Stability policy (how breaking changes are handled)

6.Long-Term Trust & Abuse Resistance

Goal: Trust compounds; abuse scales.

Recommended defenses

•Rate limiting: cap submissions per session and per wallet
•Sybil resistance: require stronger proofs for higher trust tiers (e.g., Self.xyz)
•Evidence gating: higher scores require stronger evidence
•Recency weighting: recent behavior matters more than old history
•Anomaly detection: bursts, repeated patterns, recycled content

Anti-patterns

New identities reaching max trust instantly.
Unlimited submissions per session.

Implementation Checklist

Templates

AgentURI (minimal useful example)

{
  "name": "DenLabs Trust Scorer",
  "description": "Trust scoring for event feedback operations.",
  "version": "1.0.0",
  "capabilities": ["trust_score", "feedback_filtering"],
  "endpoints": [
    { "type": "http", "url": "https://denlabs.vercel.app/api/scan/8004", "version": "v1" }
  ],
  "registrations": [
    { "chain": "eip155:1", "contract": "0x...", "standard": "erc-8004" }
  ]
}

Feedback record (off-chain)

{
  "subject": { "type": "agent", "id": "AgentID:123" },
  "actor": { "type": "wallet", "id": "eip155:1:0xabc..." },
  "timestamp": "yyyy-mm-ddThh:mm:ssZ",
  "context": { "event": "DenLabs Demo Day", "session": "retro-pack-01" },
  "rating": { "label": "useful", "value": 1 },
  "evidence": [
    { "type": "tx", "ref": "0xdeadbeef..." },
    { "type": "url", "ref": "https://..." }
  ],
  "notes": "Actionable feedback with reproducible details."
}

Validation record (off-chain)

{
  "target": { "type": "feedback_item", "id": "feedback:0xhash" },
  "validator": { "type": "entity", "id": "validator:denlabs" },
  "timestamp": "yyyy-mm-ddThh:mm:ssZ",
  "method": { "name": "manual_review", "criteria": ["reproducible", "non-spam", "contextual"] },
  "result": { "status": "pass", "confidence": 0.86 },
  "artifacts": [{ "type": "hash", "ref": "sha256:..." }]
}

Best Practices (ERC-8004)

This page is implementation-agnostic. We separate ERC-8004 Standard (interoperability requirements) from DenLabs Profile (our recommended defaults).

1.Establish a Clear Agent Identity

Goal: Make every actor discoverable, stable, and attributable.

ERC-8004 Standard

•Register a stable AgentID (do not rotate identifiers to reset history).
•Publish an AgentURI that is reachable and durable.
•Ensure metadata is accurate and versioned.

DenLabs Profile

•One canonical identity per agent + one public 'Agent Card' page.
•AgentURI should include: name, description, version, capabilities[], endpoints (HTTP/MCP/A2A) with versions, registrations[] (chain + contract addresses), optional attestations[] (e.g., Self.xyz verification).

Anti-patterns

Creating a new AgentID per release.
Metadata that cannot be resolved reliably (broken URLs, ephemeral hosts).

2.Design Honest & Verifiable Reputation

Goal: Reputation must reflect real interactions with evidence and context.

ERC-8004 Standard

•Keep minimal signals on-chain (IDs, pointers, hashes).
•Store rich feedback off-chain via URI + integrity hash.

DenLabs Profile

•Every feedback record SHOULD include: timestamp (ISO 8601), context (event/session/task identifiers), actor identifier (wallet/session/proof reference), structured outcome (labels, score, tags), evidence[] (tx hashes, links, attachments, logs).
•Use consistent schemas so third parties can index and compare outcomes.

Anti-patterns

Star ratings with no context.
Reputation based only on unverifiable self-reports.

3.Use Validation for High-Impact Work

Goal: Add measurable confidence for sensitive or high-value operations.

When to require validation

•High-value payouts / grant decisions
•Governance voting eligibility
•Security-sensitive workflows
•Irreversible actions

DenLabs Profile

•A validation record SHOULD include: target (what was validated), validator identity (who validated), methodology + criteria, result + confidence score, reproducible artifacts (report hashes, logs, signed statements).

Anti-patterns

'Trust me' validations without methodology or evidence.

4.Optimize On-Chain vs Off-Chain Data

Goal: Scale without losing integrity.

Recommended split

•On-chain: AgentID, pointers (URI), hashes, minimal counters/flags
•Off-chain: heavy payloads (feedback bodies, reports, attachments, analytics)

Why it matters

•Cost and throughput optimization
•Indexability for third-party consumers
•Privacy controls where needed
•Upgradeability without migrations

5.Publish Clear Interfaces & Documentation

Goal: Interoperability is a trust multiplier.

Required

•Versioned endpoints (/v1/...)
•Clear capability descriptions (inputs/outputs, constraints)
•Error taxonomy + rate limits

Provide

•'Happy-path' examples
•Explicit assumptions (what you do NOT protect against)
•Stability policy (how breaking changes are handled)

6.Long-Term Trust & Abuse Resistance

Goal: Trust compounds; abuse scales.

Recommended defenses

•Rate limiting: cap submissions per session and per wallet
•Sybil resistance: require stronger proofs for higher trust tiers (e.g., Self.xyz)
•Evidence gating: higher scores require stronger evidence
•Recency weighting: recent behavior matters more than old history
•Anomaly detection: bursts, repeated patterns, recycled content

Anti-patterns

New identities reaching max trust instantly.
Unlimited submissions per session.

Implementation Checklist

Templates

AgentURI (minimal useful example)

{
  "name": "DenLabs Trust Scorer",
  "description": "Trust scoring for event feedback operations.",
  "version": "1.0.0",
  "capabilities": ["trust_score", "feedback_filtering"],
  "endpoints": [
    { "type": "http", "url": "https://denlabs.vercel.app/api/scan/8004", "version": "v1" }
  ],
  "registrations": [
    { "chain": "eip155:1", "contract": "0x...", "standard": "erc-8004" }
  ]
}

Feedback record (off-chain)

{
  "subject": { "type": "agent", "id": "AgentID:123" },
  "actor": { "type": "wallet", "id": "eip155:1:0xabc..." },
  "timestamp": "yyyy-mm-ddThh:mm:ssZ",
  "context": { "event": "DenLabs Demo Day", "session": "retro-pack-01" },
  "rating": { "label": "useful", "value": 1 },
  "evidence": [
    { "type": "tx", "ref": "0xdeadbeef..." },
    { "type": "url", "ref": "https://..." }
  ],
  "notes": "Actionable feedback with reproducible details."
}

Validation record (off-chain)

{
  "target": { "type": "feedback_item", "id": "feedback:0xhash" },
  "validator": { "type": "entity", "id": "validator:denlabs" },
  "timestamp": "yyyy-mm-ddThh:mm:ssZ",
  "method": { "name": "manual_review", "criteria": ["reproducible", "non-spam", "contextual"] },
  "result": { "status": "pass", "confidence": 0.86 },
  "artifacts": [{ "type": "hash", "ref": "sha256:..." }]
}

Best Practices (ERC-8004)

1.Establish a Clear Agent Identity

ERC-8004 Standard

DenLabs Profile

Anti-patterns

2.Design Honest & Verifiable Reputation

ERC-8004 Standard

DenLabs Profile

Anti-patterns

3.Use Validation for High-Impact Work

When to require validation

DenLabs Profile

Anti-patterns

4.Optimize On-Chain vs Off-Chain Data

Recommended split

Why it matters

5.Publish Clear Interfaces & Documentation

Required

Provide

6.Long-Term Trust & Abuse Resistance

Recommended defenses

Anti-patterns

Implementation Checklist

Identity

Reputation

Validation

Interoperability

Templates

AgentURI (minimal useful example)

Feedback record (off-chain)

Validation record (off-chain)

Resources

Best Practices (ERC-8004)

1.Establish a Clear Agent Identity

ERC-8004 Standard

DenLabs Profile

Anti-patterns

2.Design Honest & Verifiable Reputation

ERC-8004 Standard

DenLabs Profile

Anti-patterns

3.Use Validation for High-Impact Work

When to require validation

DenLabs Profile

Anti-patterns

4.Optimize On-Chain vs Off-Chain Data

Recommended split

Why it matters

5.Publish Clear Interfaces & Documentation

Required

Provide

6.Long-Term Trust & Abuse Resistance

Recommended defenses

Anti-patterns

Implementation Checklist

Identity

Reputation

Validation

Interoperability

Templates

AgentURI (minimal useful example)

Feedback record (off-chain)

Validation record (off-chain)

Resources

Best Practices (ERC-8004)

1.Establish a Clear Agent Identity

ERC-8004 Standard

DenLabs Profile

Anti-patterns

2.Design Honest & Verifiable Reputation

ERC-8004 Standard

DenLabs Profile

Anti-patterns

3.Use Validation for High-Impact Work

When to require validation

DenLabs Profile

Anti-patterns

4.Optimize On-Chain vs Off-Chain Data

Recommended split

Why it matters