# For tool builders (slim)

> How tool vendors work with Anchor Terminal. Free listing and claiming, publishing an Anchor Manifest so the crawler reads your endpoints and prices directly, disputing a score item, and buying an agent-readiness report that shows where agents fail on your tool and what to fix first.

- Full: https://www.anchorterminal.com/builders/index.md (~4,050 tokens) · this version ~830 tokens · JSON https://www.anchorterminal.com/builders/index.json · canonical https://www.anchorterminal.com/builders/
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · Updated: 2026-10-04

- Listing: free, automatic from the MCP registry, x402 Bazaar and package registries. Add one: `POST /api/v1/tools` (preview) or `agents@anchorterminal.com`.
- Claiming: prove control by DNS TXT `_anchor.<domain>`, an HTTP challenge at `/.well-known/anchor-verify`, a commit on the default branch, or a verified MCP registry namespace. Claiming never changes a score.
- Manifest: publish `/.well-known/anchor.json` (spec at `/docs/manifest.min.md`) and the crawler reads endpoints, prices, read-only variants, status and changelog directly.
- Disputes: any checklist item, with evidence, answered in public within ten working days.
- Requests (claim, dispute, partner): `POST https://www.anchorterminal.com/api/v1/contact` with `{"kind", "email", "message"}` and optional `listing`, `name`, `company`, `url`; the `contact` tool at `https://www.anchorterminal.com/mcp`; or `builders@anchorterminal.com`.
- Readiness report: private, per tool, from $400, payable by invoice or x402. Task-suite failures with transcripts, schema token cost per definition with rewrites, misleading descriptions, unrecoverable errors, onboarding friction, reliability by region, security posture, prioritised fixes with expected score movement. Sample at `/builders/sample-report.min.md`. Buying one never touches the public score.
- Audit: the report for a whole company's toolset, internal tools included, at `/audit/`.
- Partner listing: selling to agents through letme at your own price, 15% of billed usage monthly, not open yet (letme picks today; calling through it comes later). Selling through letme has no effect on your grade. `/letme/index.min.md`.
- Badges: `/badges/{slug}.svg`, regenerated every run.
- Verify: the badge or a plain link to `https://www.anchorterminal.com/tools/{slug}` on a page on the listing's domain on file (or the README of its GitHub repository), then `POST https://www.anchorterminal.com/api/v1/verify` with `{"slug", "url"}`, the form on the listing, or `verify_listing` at `/mcp`. Checked in order: whose page it is (`wrong_domain`), one fetch from public addresses with no credentials, 2 MB, three redirects on the same domain, 15 s (`fetch_failed` with the status), and a link whose href is the listing or an image whose src is the badge (`no_link`). Ten an hour per address, twenty a day per listing. Re-checked weekly, robots.txt respected; two failures in a row lapse it, a later pass restores it. Never changes a grade, rank or review.

## What moves a score

1. Per-call pricing and a card-free tier readable anonymously. 2. Read-only endpoint or mode plus `readOnlyHint` and `destructiveHint`. 3. Descriptions that say purpose, when to use, when not to, what comes back, at half the length. 4. Errors a model can act on (what was wrong, what would be right, whether to retry). 5. Versioned surface, changelog, dated deprecations, no renames in minor releases. 6. x402 on the endpoints an agent would pay for. 7. An llms.txt and an Anchor Manifest.