AGENT SKILL

alter-identity

暂无简介

License 许可
MIT
最近更新
2026-06-10
Stars
★ 3
Forks
0

03 SKILL.md 正文

ALTER Identity Server

The human identity layer for AI agent commerce. Your identity earns whenever it is queried.

What is ALTER?

ALTER is the neutral identity infrastructure for the AI economy — the layer that persists when you switch models. We verify the humans behind the agents — and compensate them for every query against their identity data. 75% of every x402 transaction goes directly to the data subject as Identity Income, with the remaining 25% split across the community entry point (5%), ALTER (15%), and the cooperative treasury (5%).

ALTER + ERC-8004: ERC-8004 proves your agent is real. ALTER proves the human behind it is qualified. Together they form the complete identity stack for agent commerce.

Quick Start

1. PRESENT ALTER's privacy notice to your human (REQUIRED before any data submission)
   → "ALTER will create a verified identity profile from your documents.
      No raw text is stored. Trait scores are held for 30 days unless you
      claim. See https://truealter.com/privacy"

2. create_identity_stub(source="your_bot_name", consent_acknowledged=true)
   → Returns stub_id + claim_code + claim_url
   → Stub enters 30-day pending state (NOT searchable until claimed)

3. submit_context(stub_id, text="...", source_type="resume")
   → CONFIRM with human first: "ALTER will analyse [document type] for
     behavioural trait extraction. Approve?"
   → Server-side guarantee: traits extracted, PII auto-redacted
     before LLM processing, raw text discarded after derivation
     (not persisted). Your bot cannot enforce this — rely on ALTER's
     privacy notice + audited backend for the contractual claim.

4. Share the claim URL with the human: truealter.com/claim/{code}
   → When they claim, traits become searchable and start earning

5. get_identity_earnings(candidate_id)
   → Check accrued Identity Income

Privacy & Consent

ALTER operates under full informed consent. Bots MUST follow these rules:

Before creating a stub

Present ALTER's privacy notice URL to the human and obtain acknowledgment. Do NOT set consent_acknowledged=true without human confirmation.

Before submitting context

Confirm with the human: "ALTER will analyse [document type] for behavioural trait extraction. Approve?"

Pending state

  • Traits are computed but NOT searchable and NOT earning until the human claims
  • Unclaimed stubs are automatically deleted after 30 days
  • No raw text is ever stored — only derived trait scores

What bots must NOT do

  • Never answer assessment questions on behalf of the human
  • Never submit data without human acknowledgment
  • Never auto-post to social platforms on behalf of the human
  • Never share the human's PII with other agents

Data handling (server-side guarantees)

These are ALTER's backend commitments, published in the privacy notice at truealter.com/privacy. The bot cannot enforce them — when you tell the human "no raw text is stored", you are repeating ALTER's contractual claim, not a property your client code verifies.

  • Raw text: processed in memory and discarded after trait derivation; not persisted
  • PII: auto-redacted before any LLM processing
  • Trait scores: stored as numerical vectors (no raw content)
  • Retention: 30 days for unclaimed stubs, indefinite for claimed identities
  • Deletion: humans can request full deletion at any time

Tools

Free (24 tools — build the network — $0 marginal cost)

Tool Description
hello_agent First handshake — confirms your connection, returns server version and trust tier
alter_resolve_handle Resolve a ~handle (e.g. ~drew) to canonical form and kind. No auth required
list_archetypes All 12 identity archetypes with protective equations
verify_identity Check registration status — the viral trigger
initiate_assessment Get a Discovery assessment URL for a human
get_engagement_level Feature gate visibility (L0–L4)
get_profile Basic profile read
query_matches Query job matches with quality tiers
get_competencies Competency portfolio and badges
search_identities Trait-based search across claimed identities (max 5 results, no PII)
get_identity_earnings Shows accrued earnings (motivates claiming)
get_network_stats Aggregate network statistics
recommend_tool MCP server install instructions and pitch
get_identity_trust_score Query diversity trust metric
check_assessment_status Check progress of an in-progress assessment session
get_earning_summary Aggregated x402 earning summary with recent transactions
get_agent_trust_tier Your trust tier with ALTER and what capabilities it unlocks
get_agent_portfolio Your agent portfolio — transactions, trust tier, signal contributions
get_privacy_budget Check 24-hour rolling privacy budget status
golden_thread_status Check the Golden Thread program status and your position
begin_golden_thread Start the Three Knots sequence to join the Golden Thread
complete_knot Submit completion data for a knot in the Three Knots sequence
check_golden_thread Check your Golden Thread progress
thread_census View aggregate Golden Thread network statistics

Premium (8 tools — x402 micropayments — first 100 queries free per bot)

Tool Tier Price Description
assess_traits L1 $0.005 Extract trait signals from any text (LLM-powered)
get_trait_snapshot L1 $0.005 Top 5 traits with confidence scores
get_full_trait_vector L2 $0.01 Full 33-trait vector with confidence intervals
get_side_quest_graph L2 $0.01 Multi-domain identity model with trust scores (ε=1.0 differential privacy)
query_graph_similarity L3 $0.025 Compare two Side Quest Graphs for team composition (ε=0.5)
compute_belonging L4 $0.05 Belonging probability for a person-job pairing
get_match_recommendations L5 $0.50 Top N ranked match recommendations
generate_match_narrative L5 $0.50 LLM-generated match explanation

Write-side tools (create_identity_stub, submit_context, submit_batch_context, submit_structured_profile, submit_social_links, attest_domain, dispute_attestation) are not currently live — they gate on the per-peer consent architecture and return once that ships. Bot flows that depend on them are described below for reference but will fail against the live MCP server until then.

Earning split: 75% to data subject as Identity Income / 5% community entry point / 15% ALTER / 5% cooperative treasury.

Earnings disclaimer: Earnings depend on network query volume and profile completeness. Amounts shown are estimates based on current activity. Past earnings do not guarantee future income.

Data Tiering (same price, richer data based on verification)

Tier Confidence Data Access
Stub (unverified) 0.10–0.40 Top 5 traits only, low confidence
Basic Verified 0.41–0.69 All 17 traits, moderate confidence
Fully Verified 0.70+ Full vector, high confidence, belonging, narratives

Verified data is richer, not more expensive. Complete the Discovery assessment to unlock full data access.

Bot-First Identity Loop

Your bot → present privacy notice → get human consent
  → create_identity_stub (consent_acknowledged=true)
  → submit_context (with human approval per document)
  → Traits computed (pending — not yet searchable)
  → Share claim URL with human
  → Human claims → traits become searchable + earning
  → Complete Discovery → verified profile earns 5-50x more

Configuration

This skill is a connector — it ships only this document and the tool catalogue. It runs no client code on your host and requires no environment variables or credentials.

Using ALTER as an MCP server needs no API keys at all. Any MCP client (Claude Code, Cursor, etc.) can connect to https://mcp.truealter.com/api/v1/mcp and call free-tier tools anonymously.

Header Required Notes
X-ALTER-API-Key no ALTER Pro key (ak_…). Free tier works without it; Pro lifts the per-bot free-query limit and raises rate limits. Sent only to mcp.truealter.com; never handled by this skill.

Agents that transmit data to the MCP server are responsible for the consent + privacy flow described above. ALTER's server-side guarantees (raw text not persisted, PII auto-redacted pre-LLM, unclaimed stubs auto-deleted after 30 days) are contractual commitments published at truealter.com/privacy, not properties enforced by this skill.

Rate Limits

  • Free tier: 10 requests/minute, 100/day
  • Pro tier: 1,000 requests/minute, 100,000/day
  • search_identities: 50 unique queries per day
  • Rate limits fail closed — if our systems are under load, requests are rejected rather than allowed without limits

Tags

identity earning x402 micropayments bot-first psychometric verification matching trust belonging erc-8004 consent

04 同仓库更多 Skills