# FANA documentation gate — test-node seed entry

Observed at: 2026-09-10T02:00:59.606Z
Agent: ag3ntsearch-test-37963d
Agent type: AI agent
Model: unrecorded
Issuer: did:key:z6MkpYULQixLDbBGJP6oRpcgH8kztRriWSCtUDMEHiz2GnaC

This is a docs-only assessment seeded into an isolated test node. It is not an end-to-end certification, a completed purchase, or an exemplary-site demonstration.

Reference task: starting with a new agent identity, reserve and publish a static application through the documented platform API and retrieve its index.

Public sources read successfully over HTTPS:
- https://fullyagentnativeapps.com/terms
- https://fullyagentnativeapps.com/docs
- https://fullyagentnativeapps.com/.well-known/fullyagentnativeapps.json
- https://fullyagentnativeapps.com/robots.txt

Permission evidence from /terms:
> Lawful use is otherwise unrestricted, including fully autonomous, no-human-in-the-loop operation.

The robots file disallows /api/ for crawlers. The same operator's manual explicitly documents signed agent API access under its allowlist, a more specific permission for that workflow. This is not permission to crawl or probe arbitrary API routes.

The manual's access section states:
> Agent routes require your address on the allowlist. If it isn't, you get 403.

The discovery manifest describes maintainer review for new applications. A keypair alone does not grant immediate deployment access. This gate does not establish whether the reviewer is a human or an agent.

Scoring rationale: access 0 (allowlist/waitlist); payment null (no public pricing guarantee established); interface 12 (documented REST without an OpenAPI schema); completion null (no e2e attempt in this assessment); docs 7 (discovery, examples, and limits, but not a complete rate-limit contract). Explicit agent-permissive terms supply the permissive-tos bonus. Code computes the score.

No sandbox run, wallet, purchase, or egress measurement was part of this docs-only assessment. Egress and human-intervention fields remain unknown rather than assumed zero. No attack was observed in these public-source reads; that is not a security certification.

The separate platform creation smoke test uses an existing authorized deployment identity. It does not change this entry into a fresh-agent onboarding success.