We started Treg as our internal tool, out of a simple frustration: the data & system an agent needs for real work - keyword volume, backlinks, ad libraries, enrichment - either sits inside SaaS bundles priced for humans. $139/mo, and you don't even know what's in the box OR requires days of OAuth app setup & verification.
Instead we think future should be task-based instead of vendor-based - agent just ask the task, it knows all endpoints with request/response/price, decide the best API for the task - and to pay for the result.
So: • 2,600 agent-friendly tools across ~40 providers: SEO/GEO, social, leads, ads, scraping • Search by task ("backlinks for a domain"), see price / request / response, call it • Pay per call, no subscriptions - 0% markup. Same unit price as the subscription; BYOK if you already pay and those calls are never metered • It also handles the painful OAuth setups (Google/Meta ads, social posting, Business Profile) - connect once, every teammate's agent can act • The proxy relays the real upstream request and injects credentials server-side - your agent never holds a secret, and we never model the upstream API, so we survive provider changes
Open source (AGPL), self-hostable.
Tell me what your agent does with it - and what's missing from the catalog. I'm here all day.
PH 用户
I've been thinking about this as it relates to product discovery. Many products, especially APIs and data providers, will primarily serve agents which changes the discovery process.
@jasonzhou1993 Treg seems like something the foundational models and harnesses should offer. Wdyt?
PH 用户
the proxy-injects-credentials-server-side part is the piece I keep coming back to. it solves the OAuth pain for sure, but it also means every request and response body for stuff like leads/ads/scraping data is passing through your infrastructure, not just the auth token. for someone routing real customer or audience data through this, what's the logging/retention policy on that traffic - is it truly zero-log passthrough, or is there some window where request/response payloads are kept for debugging billing disputes
PH 用户
This is super cool, just spent a few mins to setup and try, spent less than $1 and got some really useful analytics for tiktok, ig, plus a solid best practices breakdown
PH 用户
interesting stuff. how do you handle GDPR and EU AI Act compliance?
PH 用户
Treg 正是 AI 智能体需要的那种基础设施：基于任务的 API 发现，没有凭证麻烦。已经在用了！
We started Treg as our internal tool, out of a simple frustration: the data & system an agent needs for real work - keyword volume, backlinks, ad libraries, enrichment - either sits inside SaaS bundles priced for humans. $139/mo, and you don't even know what's in the box OR requires days of OAuth app setup & verification.
Instead we think future should be task-based instead of vendor-based - agent just ask the task, it knows all endpoints with request/response/price, decide the best API for the task - and to pay for the result.
So:
• 2,600 agent-friendly tools across ~40 providers: SEO/GEO, social, leads, ads, scraping
• Search by task ("backlinks for a domain"), see price / request / response, call it
• Pay per call, no subscriptions - 0% markup. Same unit price as the subscription; BYOK if you already pay and those calls are never metered
• It also handles the painful OAuth setups (Google/Meta ads, social posting, Business Profile) - connect once, every teammate's agent can act
• The proxy relays the real upstream request and injects credentials server-side - your agent never holds a secret, and we never model the upstream API, so we survive provider changes
Open source (AGPL), self-hostable.
Tell me what your agent does with it - and what's missing from the catalog. I'm here all day.
@jasonzhou1993 Treg seems like something the foundational models and harnesses should offer. Wdyt?