Built to be understood by machines
ASHR.work exposes clean, anonymous discovery surfaces so AI assistants and agents can answer questions about the product accurately — 44 modules, 37available today. No tenant or employee data is ever exposed without credentials — the MCP server below is the one surface that goes further, reading and acting on a connected tenant's own data once it presents a scoped key.
Capabilities manifest
Machine-readable JSON: product, modules, statuses, surfaces (schema demystify-capabilities/v1).
/api/capabilitiesllms.txt
Concise index of key pages and help articles.
/llms.txtllms-full.txt
Full corpus: every module with status, the FAQ, and the help center.
/llms-full.txtMCP server
A Model Context Protocol server: discover modules and journeys anonymously (list/search/describe, get_user_journeys), or connect an API key/OAuth client to read a tenant’s own data and take real actions on it (apply for leave, create an employee, move a candidate…) — see the developer docs and the repo mcp/ directory.
/developersCrawling & trust
We deliberately allow major AI crawlers (GPTBot, OAI-SearchBot, ChatGPT-User, PerplexityBot, Perplexity-User, ClaudeBot and more) on public marketing and help content, and disallow the authenticated app. See robots.txt. Everything on this page is public product metadata — accurate by construction, generated from one source.