Pactum Emerging

Name: Agent Native Offer Audit — Pactum
Item: Pactum
Rating: 72
Author: Agent Native Offers

AUDIE Score: 72/100 · Audited 2026-04-10 · Website: https://pactum.cc · Machine-readable: JSON

Pillar Scores

P1 Signal Architecture — 16/25

P2 Clarity Stack — 20/25

P3 Trust Envelope — 12/20

P4 Velocity Triggers — 9/10

P5 Gravity Design — 15/20

Executive Summary

Pactum sits solidly in the **Emerging** tier (72/100) — the highest-scoring company audited to date — thanks to an API-first architecture that genuinely enables agent-to-agent commerce, clean semantic precision in its offer language, and real on-chain trust signals. Its strongest pillar is Velocity (9/10): an agent can discover, evaluate, and purchase with zero human involvement, which is the correct design for agent-native infrastructure. The critical gap is Signal Architecture: despite being purpose-built for AI agents, Pactum has no schema.org markup and no /llms.txt — making it invisible to agents who discover services via AI retrieval systems rather than direct API calls. Publishing /llms.txt and adding Offer schema markup are two low-effort actions that could push Pactum into Agent-Ready territory (85+) within days.

Strongest Signals

API-first architecture (P1-D: 5/5) Every core action — discovery, evaluation, purchase — is API-callable without human involvement. This is the gold standard for agent-native velocity.
Semantic precision (P2-E: 5/5) USDC pricing is explicit, terms are precise, and there are no vague marketing claims. Agents can parse Pactum's offers without ambiguity.
On-chain audit trail (P3-C: 4/5) Blockchain settlement provides an immutable, agent-verifiable transaction record — a rare trust signal in early agent infrastructure.
Network effects (P5-D: 5/5) The marketplace compounding model is genuine; value increases with each new provider and buyer agent.

Critical Gaps

No schema.org markup (P1-A: 2/5) Agents relying on structured data signals will find nothing machine-tagged on Pactum's pages. This blocks discovery via AI retrieval systems.
No /llms.txt (P1-C: 2/5) No agent identity document exists. Agents have no standardized way to understand what Pactum is, what it offers, and how to interact with it programmatically.
No verifiable performance data (P3-A: 2/5) Zero uptime SLA, no status page, no third-party reliability verification. For autonomous agents making purchasing decisions, this is a significant trust gap.
No agent memory layer (P5-B: 2/5) Each interaction starts from scratch. Agents cannot build on prior transaction history, preferences, or usage patterns — limiting the depth of agent loyalty.

Priority Actions

Publish /llms.txt — +5 pts · P1 · Effort: Low
Add schema.org/Offer markup to marketplace listings — +4 pts · P1 · Effort: Low
Publish a public status page with uptime SLA — +4 pts · P3 · Effort: Low
Build an agent memory / account history API — +5 pts · P5 · Effort: Med
Document explicit substitution and fallback rules — +3 pts · P2 · Effort: Low

All 20 Criteria

P1-A Structured Data — 2/5
Homepage HTML has minimal schema.org markup; no Offer, Product, or AggregateRating schema detected. Basic Organization markup present but not sufficient for agent parsing.

P1-B Machine-Readable Pricing — 3/5
Fee structure discoverable via page scraping in HTML text; USDC amounts are explicitly stated in prose/tables but not tagged with schema.org/Offer or structured JSON. Parseable but not machine-tagged.

P1-C llms.txt / Agent Layer — 2/5
No /llms.txt found at root. Developer documentation exists and is technically oriented, but not explicitly structured for LLM or agent consumption.

P1-D API / MCP Availability — 5/5
Full REST API with documentation; agent-callable endpoints for discovering, evaluating, and purchasing marketplace services. API-first architecture clearly designed for programmatic consumption.

P1-E Discoverability (GEO) — 4/5
Strong developer-focused content with clear technical narrative. Indexed content explains agent-to-agent commerce well; structured for technical audience retrieval.

P2-A Offer Completeness — 4/5
Marketplace listings include service description, USDC pricing, and API endpoint in a consistent format. Nearly single-source parseable; minor gaps in edge-case coverage.

P2-B Scope & Limits — 4/5
API rate limits documented in developer docs; per-listing service caps stated. Not structured as machine-tagged JSON but clearly present and findable.

P2-C Substitution Rules — 3/5
Some fallback guidance present in FAQ and docs; no explicit, machine-readable substitution protocol for failed or unavailable transactions.

P2-D Conditional Logic — 4/5
Pricing conditions are visible and documented; vendor-specific conditions accessible via listing detail API. Some conditions require parsing multiple pages.

P2-E Semantic Precision — 5/5
Precise technical language throughout; USDC amounts explicitly stated; no vague marketing claims detected ("best-in-class," "powerful," etc.). Definitions consistent across pages.

P3-A Verifiable Performance — 2/5
No uptime SLA, status page, or third-party performance verification found. Self-reported reliability claims only; no G2, Trustpilot, or independent audit data.

P3-B Scoped Permissions — 3/5
API key scoping available with some permission tiers documented. No explicit agent-time-bounded or action-bounded permission scopes found.

P3-C Audit Trail — 4/5
On-chain USDC transaction records provide immutable, verifiable audit trail on Base blockchain. Not human-readable dashboard but blockchain-verifiable for agent systems.

P3-D Behavioral Consistency — 3/5
Marketplace terms versioned; change notification policy referenced in docs. Platform is relatively new; long-term stability track record not yet established.

P4-A Friction-Free Activation — 5/5
Instant API key issuance; no sales call, approval gate, or human review required. Fully self-serve onboarding; agents can activate and transact without human involvement.

P4-B Agent Decision Signals — 4/5
Free trial credits available; marketplace discovery API allows agent to evaluate services before purchase; pricing signals clearly communicated. Missing: explicit "recommended for agents" or confidence signals.

P5-A Integration Depth — 3/5
USDC wallet integration and transaction history create moderate switching cost. API integration is meaningful but not deeply proprietary; marketplace data could migrate.

P5-B Agent Memory Layer — 2/5
No persistent agent memory or personalization layer found. Each transaction is independent; no account-level history surfaced via API for agent consumption.

P5-C Programmatic Renewal — 5/5
Subscription and recurring payment endpoints fully documented and agent-callable. Auto-renewal supported natively via API without human gate.

P5-D Compounding Value — 5/5
Network effects compound clearly: more service providers → more agent buyers → more providers. 100+ listings create genuine compounding discovery value; marketplace grows more valuable as agents transact.

Rubric v1 (April 2026). Scores reflect the company's state on the audit date and may have improved since.