You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The company registry (registry/*.json) is currently populated by manual PRs. Users of ats-index will naturally discover companies on supported ATS platforms as a side-effect of their searches — companies that could be registered but aren't.
What if we made it easy for users to contribute back what they discover?
The design tension
Too low friction
Too high friction
Auto-submit every slug that worked → registry fills with typos, one-off uses, throwaway slugs
PR to the repo → nobody bothers
The right path is probably in the middle: pre-validated submission, one command.
Phased proposal
Phase 1 — CLI command
ats-index contribute --slug whoop --ats lever --name WHOOP --sector wearables
The tool auto-verifies the slug against the ATS API before accepting it. If valid, it opens a pre-filled GitHub issue (or creates it directly if gh is available).
Phase 2 — Passive hint
When fetch <company> succeeds on a slug not in the registry, the CLI prints:
Not yet in registry — share it: ats-index contribute --slug X --ats Y
AI assistants using ats-index via MCP suggest contributions conversationally when they encounter a miss or discover a new valid slug.
Phase 4 — Web form / service (later)
A public submission form that validates via the same API logic and commits to a review branch.
Privacy principle
Only the infrastructure facts travel upstream: slug, name, sector. Never the user's query, target role, or personal context. Keep a firm line between "community contribution to shared infrastructure" and "collection of user behavior data."
Open questions
Is Phase 1 (CLI contribute) enough on its own, or does Phase 2 (passive hint) add real value?
How should we prevent low-quality submissions? Validation-by-API is the first defense — what else?
For Phase 3 (MCP), should the AI ask for user consent per submission, or batch-suggest at end of session?
Should submissions include a verifiedAt timestamp so the registry can self-prune stale entries?
Anything else worth considering before drafting a concrete design?
This is a direction, not a plan. Curious what others think before committing to a shape.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
-
The idea
The company registry (
registry/*.json) is currently populated by manual PRs. Users of ats-index will naturally discover companies on supported ATS platforms as a side-effect of their searches — companies that could be registered but aren't.What if we made it easy for users to contribute back what they discover?
The design tension
The right path is probably in the middle: pre-validated submission, one command.
Phased proposal
Phase 1 — CLI command
The tool auto-verifies the slug against the ATS API before accepting it. If valid, it opens a pre-filled GitHub issue (or creates it directly if
ghis available).Phase 2 — Passive hint
When
fetch <company>succeeds on a slug not in the registry, the CLI prints:Phase 3 — MCP-native (once #1 ships)
AI assistants using ats-index via MCP suggest contributions conversationally when they encounter a miss or discover a new valid slug.
Phase 4 — Web form / service (later)
A public submission form that validates via the same API logic and commits to a review branch.
Privacy principle
Only the infrastructure facts travel upstream: slug, name, sector. Never the user's query, target role, or personal context. Keep a firm line between "community contribution to shared infrastructure" and "collection of user behavior data."
Open questions
contribute) enough on its own, or does Phase 2 (passive hint) add real value?verifiedAttimestamp so the registry can self-prune stale entries?This is a direction, not a plan. Curious what others think before committing to a shape.
Beta Was this translation helpful? Give feedback.
All reactions