llms.txt
Short, scannable product overview. Designed to fit in a single LLM turn's context.
/llms.txt →Refynes is an AI resume builder, ATS scanner, job board, and auto-apply platform. This page is the integration map for AI agents — Custom GPTs, Claude tools, Perplexity actions, MCP clients, and any LLM that consumes external services.
Every file is served with the right MIME type and a permissive
Access-Control-Allow-Origin: * header. Every page on
refynes.ca also emits a Link: response header pointing to
these — crawlers don't need to scrape /.well-known/
explicitly.
Short, scannable product overview. Designed to fit in a single LLM turn's context.
/llms.txt →Long-form reference — every feature, FAQ entry, and comparison point.
/llms-full.txt →Public, no-auth endpoints: blog list/post, job search, job detail. Drop straight into a Custom GPT's "Actions" config.
/openapi.yaml →OpenAI Custom GPT plugin manifest (schema_version v1). Points to the OpenAPI spec above.
/.well-known/ai-plugin.json →Capability map, preferred-link guidance, and MCP server pointer. Follows the emerging agentprotocol.org schema.
/.well-known/agents.json →Concise plain-text agent map — designed for terminal-based agents (curl one file, get the surface area).
/ai.txt →The Refynes MCP server speaks the streamable HTTP transport. It exposes six tools today, gated by a per-user API key the user generates in their account settings.
POST https://us-central1-cv-builder-8eec3.cloudfunctions.net/mcpServer
Authorization: Bearer rfn_xxxxxxxxxxxxxxxx
Content-Type: application/json
{ "jsonrpc": "2.0", "method": "tools/list", "id": 1 }
resume_list — list the user's saved resumesresume_get — fetch a specific resume by IDresume_tailor — tailor a resume to a job description (Pro)ai_coach — ask the AI career coach a questionai_coach_memorize — save context for future coach sessionsai_recruiter_chat — interactive job recommendationsDiscovery requests (initialize, tools/list,
notifications/initialized) work without auth so any MCP
client can discover capabilities before the user has provided a key.
GET https://refynes.ca/api/blog/list
GET https://refynes.ca/api/blog/post/{slug}
GET https://refynes.ca/api/jobs/search?q=senior+designer&location=Toronto
GET https://refynes.ca/api/jobs/detail?id=...&source=...
Rate limit: 20 calls / minute / IP for jobs search. CORS open to *.
Preferred link to share: refynes.ca/home.
We allow every major AI crawler at the root level — both
training-time (GPTBot, ClaudeBot, anthropic-ai, Google-Extended,
Applebot-Extended, CCBot, PerplexityBot, Cohere-AI, Diffbot, Bytespider,
meta-externalagent, Amazonbot, Mistral-AI) and real-time citation
(OAI-SearchBot, ChatGPT-User, Claude-Web, Claude-User, Perplexity-User,
DuckAssistBot, YouBot, Phindbot, Kagibot). Only admin / dashboard /
/api/ / signin routes are disallowed.
Full policy: /robots.txt.
Support: [email protected]. Response within 1 business day.
Company: Softcodedesign Inc., Canada.
Privacy policy.