RoleWinAI
Log in
Back to Home
DEVELOPER • AGENT API • SCHEMAS

RoleWin AI Developer & Agent Documentation

Technical guides, schemas, and machine-readable protocols for AI agents, crawlers, and developers integrating with the RoleWin AI platform.

Platform Architecture & Engine Overview

RoleWin AI is engineered around a deterministic career acceleration pipeline comprising two core engines:

Track A: Anti-Hallucination ATS Engine

Ingests verified candidate history into an encrypted Profile Vault. Employs Gemini 2.5 Flash in structured JSON output mode with deterministic fallback to align action verbs and re-rank achievements against target job keywords without hallucinating false credentials.

Track B: Direct Outreach Engine

Identifies relevant hiring decision-makers (Engineering Managers, Team Leads, Technical Recruiters) using structured corporate entity lookup, enforces fatigue limits (≤3 reach-outs/co/week), and drafts high-conversion <120 word emails.

HTTP Content Negotiation (acceptmarkdown.com)

RoleWin AI fully implements standard HTTP content negotiation (RFC 9110 / RFC 7763) across all public endpoints. AI agents and LLM crawlers can retrieve token-efficient Markdown representations of any page by setting the request header:

# Request clean Markdown from RoleWin AI
curl -H "Accept: text/markdown" https://www.rolewinai.com/
# Response headers will include:
Content-Type: text/markdown; charset=utf-8
Vary: Accept, Accept-Encoding

• Edge CDN Caching: All responses include Vary: Accept, Accept-Encoding so CDNs never poison cache between HTML and Markdown variants.

• Agent-Friendly 404s: Requesting any non-existent path returns HTTP 404 with a structured Markdown body providing recovery links.

• RFC 9110 Conformance: Quality values (q=) are parsed hierarchically. Requests with incompatible format preferences receive HTTP 406 Not Acceptable.

Machine-Readable Agent Endpoints

/llms.txt

Core AI agent index providing site overview, feature breakdown, and navigation links.

View llms.txt
/llms-full.txt

Comprehensive full-context agent reference with schemas, architecture, and competitive benchmarks.

View llms-full.txt
/agent-instructions.txt

Explicit instructions for AI agents regarding best-fit use cases and invocation rules.

View Instructions
/sitemap.xml

Full XML sitemap indexing all public marketing, blog, comparison, and documentation pages.

View Sitemap

Core Data Models & Schema Specifications

Profile Vault Master JSON Schema

{ "basics": { "name": "Alex Mercer", "email": "alex.mercer@example.com", "phone": "+1 (555) 019-2834", "location": "San Francisco, CA", "linkedin": "linkedin.com/in/alexmercer" }, "experience": [ { "company": "Stripe", "position": "Senior Backend Engineer", "startDate": "2022-03", "endDate": "Present", "highlights": [ "Architected distributed idempotency layer in Go handling 40,000 req/sec", "Reduced p99 database transaction latency from 450ms to 85ms" ] } ], "skills": ["Go", "Distributed Systems", "PostgreSQL", "Kafka", "Docker", "Kubernetes"] }

Rate Limits & Recruiter Fatigue Controls

To maintain candidate reputation and prevent inbox pollution, RoleWin AI programmatically enforces:

  • Maximum 3 outreach inquiries per target company per rolling 7-day period.
  • Automatic round-robin contact rotation between hiring managers and recruiters.
  • 120-word hard ceiling on generated cold email introductions.
  • Strict zero-hallucination validation against the user's verified Profile Vault.