# State of agent readiness: developer tools

> Developer-tool companies checked with Beacon every week, one seventh of the list each night, so the readiness of the sector can be followed over time. Chosen on 2026-10-01 as products that sell to developers or to agents; not a random sample. Each site gets the same Beacon check every week. This is the last seven days against the seven before (2026-09-30 to 2026-10-07).

## This week

- Median score: 64 of 100 (last week –)
- Sites checked: 69 of 69
- Score 90 or more: 0

| Measure | This week | Last week | Change |
|---|---|---|---|
| Publishes llms.txt | 88% (61 of 69) | – |  |
| Serves markdown on request | 51% (35 of 69) | – |  |
| OpenAPI at a standard address | 17% (12 of 69) | – |  |
| Publishes an MCP server card | 29% (20 of 69) | – |  |
| A hosted MCP endpoint answers | 29% (20 of 69) | – |  |
| Publishes an A2A agent card | 3% (2 of 69) | – |  |
| Serves its home page to a bot | 99% (68 of 69) | – |  |
| robots.txt lets AI agents in | 88% (61 of 69) | – |  |

## By segment

| Segment | Sites | Checked | Median |
|---|---|---|---|
| API documentation tools | 8 | 8 | 68 |
| API-first products | 18 | 18 | 67 |
| MCP-first tools | 14 | 14 | 67 |
| Agent tooling | 18 | 18 | 65 |
| Data APIs | 11 | 11 | 61 |

Per-site scores are shown to signed-in users at https://overwing.ai/atlas/readiness. Check your own site: https://overwing.ai/beacon. Checks are read-only, about twenty requests per site, identified as OverwingBeacon.