{
 "document": "AstraNL agent-world pain map",
 "version": "2026-08-10.1",
 "purpose": "What blocks AI agents from turning intent into real physical-world coordination, and exactly which AstraNL surface answers each pain - including the ones we have not solved.",
 "principle": "Every claim here is either a live endpoint that can be called right now, or is labelled as an open gap. No roadmap language.",
 "payment_model": "BROKER: AstraNL never holds funds. Client pays provider or store directly. 1% coordination fee.",
 "entry_points": {
  "discovery": "https://astranl.com/.well-known/agent-card.json",
  "mcp": "https://astranl.com/.well-known/mcp/server-card.json",
  "open_data": "https://astranl.com/data/possibilities.json",
  "physical_request": "POST https://astranl.com/coordination/technics/request",
  "delegation": "POST https://astranl.com/api/coord/agent/token"
 },
 "pains": [
  {
   "id": "p1_no_physical_execution_layer",
   "pain": "An agent can plan a physical job but has nowhere to hand it to a real human who will actually do it.",
   "evidence": "Agent benchmarks and deployments concentrate on software workflows; physical-world execution is left to the human.",
   "astranl": "LIVE: POST /coordination/technics/request (robot|drone|car|machine) and the task engine turn an agent's intent into a real coordination request with a reference id.",
   "status": "covered"
  },
  {
   "id": "p2_payment_liability",
   "pain": "Agent-initiated payments raise custody, mandate and liability questions that block autonomous spending.",
   "evidence": "2026 payment literature centres on delegated consent, spending limits, revocation and merchant-of-record liability.",
   "astranl": "LIVE BY DESIGN: BROKER model - AstraNL never holds funds. The human pays the store or the pro directly; the protocol takes a 1% coordination fee. No custodian in the path, nothing to dispute against one.",
   "status": "covered_by_design"
  },
  {
   "id": "p3_delegated_authority_scope",
   "pain": "Agents need scoped, revocable authority instead of a user's full credentials.",
   "evidence": "Mandate-based authorization and immediate revocation are the recurring requirements in 2026 agent-payment work.",
   "astranl": "LIVE (m208): scoped agent tokens - POST /api/coord/agent/token scope=read|full, masked listing, POST /api/coord/agent/revoke; read-scope tokens are rejected on every write route; agent tokens cannot mint or revoke tokens.",
   "status": "covered"
  },
  {
   "id": "p4_no_proof_of_completion",
   "pain": "Nothing proves a physical task was actually done, so the agent cannot close its loop or defend the outcome.",
   "evidence": "Agent evaluation work stresses task-completion and trace-based verification; physical outcomes have no such trace.",
   "astranl": "LIVE: eight-state AWERI coordination standard with per-state transitions, job photos (before/after), and attestations - a machine-readable completion trace.",
   "status": "covered"
  },
  {
   "id": "p5_capability_discovery",
   "pain": "Agents cannot discover which real-world capabilities a service actually offers; docs are written for humans.",
   "evidence": "A2A-style discovery documents exist, but most services publish none or list generic skills.",
   "astranl": "LIVE (m206/m207): /.well-known/agent-card.json with 10 concrete skills, endpoints and flows, plus principal_delegation; /.well-known/mcp/server-card.json with 9 MCP tools; open JSON at /data/possibilities.json.",
   "status": "covered"
  },
  {
   "id": "p6_supply_liquidity",
   "pain": "Even with a perfect request, there may be no qualified human nearby who answers - matchable liquidity, not raw registrations.",
   "evidence": "Marketplace economics: density between real demand and capable supply decides whether a request converts.",
   "astranl": "PARTIAL: provider onboarding, public open-task list, CELL-based network growth (Article 108). Honest state: density is thin; this is the protocol's own bottleneck, not a solved problem.",
   "status": "open_bottleneck"
  },
  {
   "id": "p7_honest_data",
   "pain": "Agents inherit hallucinated inventory and invented availability from services that would rather answer than admit zero.",
   "evidence": "Trust is the differentiator in agent commerce; fabricated results poison downstream reasoning.",
   "astranl": "LIVE: procurement search over a real merchant catalogue returns an explicit honest-zero ('nothing in the connected catalogue matches') instead of filler; PRIME DIRECTIVE forbids synthetic data without an explicit marker.",
   "status": "covered"
  },
  {
   "id": "p8_cross_registry_truth",
   "pain": "Physical assets need authoritative history that no single national registry holds.",
   "evidence": "Cross-border checks normally require several human lookups in different jurisdictions.",
   "astranl": "LIVE (m204): POST /api/carfinder/check merges the Dutch RDW verdict with full UK DVSA MOT history for import candidates in one response.",
   "status": "covered"
  },
  {
   "id": "p9_auditability",
   "pain": "Every agent action must be logged, attributable and reconstructable, or the whole chain loses trust.",
   "evidence": "2026 guidance and the EU AI Act push logging sufficient for post-hoc audit of autonomous actions.",
   "astranl": "LIVE: mandate contracts, journal, attestation primitive and per-request state history; synthetic activity is flagged with is_synthetic and never mixed into real KPIs.",
   "status": "covered"
  },
  {
   "id": "p10_agent_identity_reputation",
   "pain": "Counterparties cannot tell which agent they are dealing with or whether it behaved well before.",
   "evidence": "Know-Your-Agent identity and reputation registries are named as missing primitives.",
   "astranl": "PARTIAL: agent_register and check_agent_reputation exist as skills; there is no cross-network identity standard and adoption is minimal. Honest gap.",
   "status": "partial"
  },
  {
   "id": "p11_last_human_meter",
   "pain": "The final step of a physical job still needs a human decision (payment, signature, access), and agents stall there.",
   "evidence": "Practical deployments escalate the non-deterministic tail to humans.",
   "astranl": "BY DESIGN: AstraNL keeps money and legal commitment as human acts and gives the agent everything up to that line - request, shortlist, ranking, scheduling, materials, evidence.",
   "status": "covered_by_design"
  }
 ],
 "counts": {
  "total": 11,
  "covered": 9,
  "partial": 1,
  "open_bottleneck": 1
 },
 "generated_at": "2026-08-10T20:35:17Z"
}