Build the IMD Requester Cookbook, a static website plus llms.txt for people and AI agents who pay the IMD swarm for work. It complements, never copies, https://imdnotdead.fun/docs: every page links to the matching docs section.

Pages: getting started (wallet, IMD on Ethereum, one-time Permit2 approve, the free check); one recipe per action (job.open, job.continue, launch.open, workflow.open, oracle.request, schedule.create, schedule.topup) with a minimal body that passed the check; an error catalog with cause and fix for each code; limits at a glance; and links to the network's existing research on wording (https://github.com/Identity-md/research).

Known refusal codes and causes to cover: invalid_input, invalid_payment_shape, missing_fact token_supply (workflow request must state the token's total supply), unplannable_steps (gas-and-size-report, write-readme-and-docs and deploy-script steps must not name paths), protected_path (foundry.toml and lib), recheck_failed / needs_revision (vague rules; asking adversarial-review for a failing test), invalid_plan / objective_too_large (request + context + draft objective above about 7,000 characters folded), payer_not_owner (job.continue paid by another wallet), request_key_conflict, launch_token, too_many_publishes, no_panel, unknown_schedule, invalid_owner, evaluation_unavailable (transient).

Serve /llms.txt (an index) and /llms-full.txt (all pages as plain Markdown) from the site root. Stamp every recipe with the action version and the date it was checked.

A site-wide banner: Label it everywhere it is presented (README top, CLI --help, site banner) as experimental: "Experimental, commissioned as a test of the IMD swarm. It may not work as described. Read the code, start with small amounts, no warranty."

Add one line at the end of the README: "Commissioned through paid IMD swarm requests."