HireMind Documentation

Autonomous Hiring Intelligence Docs

Scope, contracts, and architecture for the HireMind agent capability in a concise technical reference format.

Positioning

HireMind is a pay-per-score autonomous hiring intelligence capability built for machine-to-machine ecosystems.

API Contract (MVP)

Endpoint naming in implementation may differ, but the logical contract below reflects PRD scope.

POST /score-cv

Request:
{
  "cv_text": "string",
  "job_description": "string",
  "scoring_profile": "string",
  "language": "optional",
  "seniority_hint": "optional"
}

Response:
{
  "overall_score": 0-100,
  "role_fit_score": 0-100,
  "ats_score": 0-100,
  "strengths": [],
  "red_flags": [],
  "rewrite_suggestions": [],
  "confidence_score": 0-1,
  "scoring_version": "v1.0",
  "rubric_hash": "0x...",
  "model_version": "...",
  "timestamp": "..."
}

Core Objective

HireMind is a machine-to-machine hiring intelligence capability. Agents submit CV text and job context, then receive deterministic, structured, versioned scoring output.

•API-first CV scoring capability
•Deterministic output contract per scoring version
•Micropayment-ready request flow
•Built for agent ecosystems, not consumer CV editing

MVP Scope

MVP is intentionally narrow: one capability delivered reliably. The system scores CV-to-role fit and returns auditable machine-readable output.

•Primary capability: CV scoring as a service
•Input: cv_text, job_description, scoring_profile
•Output: structured JSON with scores and reasoning
•Deterministic behavior anchored to scoring_version

Scoring Engine Design

Scoring combines parsing, rule validation, and model-based evaluation under versioned control.

•Parsing layer: structure extraction, impact statements, timeline signals
•Rule layer: ATS checks, length checks, missing contact/timeline issues
•Evaluation layer: role-fit, skill alignment, seniority consistency
•Versioning: scoring_version, rubric_hash, model_version for reproducibility

x402 Micropayment Readiness

Payment logic is middleware-oriented and separated from scoring logic so economics can evolve without destabilizing scoring behavior.

•Missing payment context can return HTTP 402 in payment-enabled mode
•Per-request micropayment baseline
•Optional volume tiers and enterprise SLA plans
•No coupling between payment middleware and scoring core

ERC-8004 Compatible Identity Binding

Capability identity and version integrity can be exposed through minimal attestable metadata without leaking candidate data.

•Agent identity reference and capability metadata
•Scoring version hash and request count telemetry
•On-chain minimum: hashes, timestamps, agent references
•No CV payload storage on-chain

Capability Registry

A discovery endpoint enables other agents and registries to integrate capability metadata programmatically.

•Suggested endpoint: GET /capability.json
•Includes schemas, cost_per_call, avg_latency, scoring_version
•Includes attestation reference for trust checks
•Designed for agent marketplace compatibility

Excluded From MVP

The MVP avoids adjacent complexity so reliability, latency, and deterministic behavior can be validated first.

•Bias detection modules
•Multi-language expansion
•Human dashboard UX
•DAO governance, token issuance, marketplace features

Phase 2 Direction

After MVP stabilization, the roadmap expands toward cross-agent performance intelligence and stronger verification loops.

•Reputation scoring and confidence calibration
•Inter-agent benchmarking primitives
•Verifiable usage attestations
•Performance comparison API