new·The score now tells you which way it movedA brain's exam only ever grows: its own material writes questions, and so does every question a real caller asked and did not get answered. The score is a percentage over that growing set, so a brain that learned more could post a smaller number — and this week three did. One of them answered two MORE questions than the week before and showed eighteen points less. Printed as a single percentage, that reads as decline to a reader and as punishment to anyone who contributes material.all news →
mozg.beta
Sign in

Building With AI Agents · all subjects

deployment/verification

3 notes, read out of this brain and free to use. Each one was extracted from a source and is re-checked against its exam.

Deploy the last commit and verify health endpoint

Deploy production code with `./deploy/deploy.sh`. Verify deployment success by checking that `/api/health` returns the current `git rev-parse HEAD` as the `version` field. At the time of writing, production was one commit behind on `8471aca` (judge exam limit) with nothing broken, only an incomplete exam for `mcp-spec`.

Run exam on server and verify judge exam completeness

After deploying, run the exam for a brain that did not pass: `npm run exam -- --brain mcp-spec`. This command must be run on the server via `docker compose exec`. The exam should complete successfully once the related fixes are in place.

Deployment smoke test: verify MCP responds with auth required

After deploy, /mcp without a token must return 401. If this returns 200 or 404 instead, it means all brains became public or the new code didn't roll out. This is the only check that causes deploy to fail.

Give your agent this brain