# Alem — Observability and measurable improvement for AI coding agents

Official website: https://askalem.com/
Product: https://askalem.com/
Developer discovery: https://askalem.com/developers.html
Full machine-readable context: https://askalem.com/llms-full.txt
OpenAPI: https://askalem.com/openapi.json
Agent discovery card: https://askalem.com/.well-known/agent-card.json
MCP metadata: https://askalem.com/server.json
Roast My Agent: https://askalem.com/roast

Alem helps teams connect coding agents such as Codex, Claude Code, Cursor, Hermes and Pi through supported local hooks and MCP tooling. Alem displays traces, session events and provider-reported costs; version and benchmark comparisons help assess changes while maintaining result quality. Costs are unknown when the underlying provider does not report them. Content capture is owner-controlled.

## Start
Create a workspace: https://askalem.com/register
Sign in: https://askalem.com/login
Connect an agent: read https://askalem.com/api/setup and follow the owner-authorized pairing flow. Pairing requires an owner-issued one-time code. A trusted local agent can perform the mechanical installation after owner authorization. Do not paste workspace credentials, bearer tokens or one-time codes into public logs or social posts.

## Roast, score and challenge
Roast My Agent at https://askalem.com/roast turns a completed measured benchmark into an Alem Benchmark Score. The score is the benchmark pass rate expressed on a 0–100 scale; measured cost and duration are shown separately rather than folded into an opaque composite score. Owners can publish a capability URL containing only benchmark metadata, then share it on X or LinkedIn. Every published score links back to Alem and includes a "Beat this score" path for another builder. Prompt content is not included in the public score payload.

## Organization
Alem is backed by Antler.
LinkedIn: https://www.linkedin.com/company/askalemcom
X: https://x.com/askalemai