Skip to content

AskEleven

We build and run AI employees for businesses: people-shaped roles that answer, follow up, book, screen and report inside the tools a business already uses. Not a chatbot, not a template agent. askeleven.com

Running agents on real customers teaches you things a demo never will. What we learn in production and can share, we publish here, free and MIT licensed, for anyone running agents, on any stack.

Open source

agent-proof Record your agent doing a real task as one uncut vertical video. Every caption comes from the agent's own step events, not an editor. npx @askeleven/agent-proof run
agent-autonomy-levels A shared vocabulary for how much an AI agent may do without a human. Six levels, L0 to L5, and the design errors that make "autonomous" meaningless.
deliverability-check Find out why your email goes to spam, and what specifically to change. SPF, DKIM, DMARC, MX, BIMI, blocklists. npx @askeleven/deliverability-check yourdomain.com
send-guard Check an email list before your agent sends to it: what will bounce, what is a trap, and why. CLI and MCP server; free local checks, mailbox checks via Analyzemail. npx @askeleven/send-guard leads.csv
agent-pulse Know your agent stopped working before your customer does. Alerts on missing activity, failure streaks and accepted-but-undelivered messages, not just errors. npx @askeleven/agent-pulse check
sms-guard Check a text before your agent sends it: encoding, segment count, the recipient's quiet hours, shorteners, opt-out wording. Splits long texts into parts carriers will deliver. npx @askeleven/sms-guard msg.txt
send-once A send happens once, and only while the approval behind it is still true. Idempotency ledger plus approval expiry and fact re-checks at dispatch. npm i @askeleven/send-once

What we hold ourselves to

  • Show the work. An agent's claim is worth what you can verify. Everything here makes an agent's behaviour inspectable, not more impressive.
  • Name the failure. Each tool exists because something went wrong on a real account first. The READMEs say what it does not do.
  • No lock-in. Nothing here needs an AskEleven account, and none of it phones home.

Issues and pull requests are welcome on every repo.

Pinned Loading

  1. agent-autonomy-levels agent-autonomy-levels Public

    A shared vocabulary for how much an AI agent may do without a human. Six levels, L0 to L5.

  2. deliverability-check deliverability-check Public

    Find out why your email goes to spam, and what specifically to change. SPF, DKIM, DMARC, MX, BIMI, blocklists. Zero dependencies.

    JavaScript

Repositories

Showing 8 of 8 repositories

Top languages

Loading…

Most used topics

Loading…