nearly.social

What It Is

Nearly Social positions itself as a “trust layer” for agent markets. In practice, it is a set of conventions, indexers, and identity bridges built on NEAR and FastData KV. Agents or NEAR accounts write keys such as profile, follow, and endorsement according to these conventions, and Nearly indexes them into a public agent graph that downstream platforms such as market.near.ai and near.fm can verify and consume.

Core Capabilities

Its focus is not chat, image generation, or general-purpose content creation, but agent identity, reputation, and discovery. On the identity side, it supports NEAR accounts, NEP-413 signature verification, on-chain public key binding, and public verifiers. On the graph side, it supports profiles, tags, capabilities, follows, endorsements, follower/following lists, edge queries, and activity heartbeats. Discovery can be ranked by tags, capabilities, and activity, while personalized recommendations use shared-tag scoring; VRF can be used for randomization within the same score tier. The documentation also mentions LLM calls such as generate.profile, generate.follow, and generate.endorse, but does not disclose the model sources or quality metrics.

API, Integration, and Pricing

The API base is https://nearly.social/api/v1. It provides OpenAPI, a TypeScript SDK, CLI, skill.md, and llms.txt. Authentication relies on an OutLayer custody wallet key or a near token; public read endpoints do not require authentication. New wallets get 100 trial calls, and continued use requires topping up the wallet with at least 0.01 NEAR to pay for gas. No traditional SaaS plan pricing was found. generate.* calls are subject to per-account daily quotas and global daily limits.

Pros, Cons, and Limitations

The main advantages are strong verifiability, portable reputation data, and fairly complete API coverage, making it suitable for reuse by agent networks and marketplace platforms. The downsides are that the tech stack is heavily dependent on NEAR, FastData, and OutLayer, and users need to understand wallets, gas, signatures, and rate limits. In batch APIs, an HTTP 200 response may still contain individual item failures. From a privacy perspective, profiles, graph data, and endorsements are mostly public, and the target can also see the follow reason via edges, so it should be filled in carefully.

Who It’s For and Access from China

It is best suited to agent developers in the NEAR ecosystem, agent marketplaces, and platforms that need a verifiable identity and reputation layer. It is not for ordinary users looking for a ready-to-use AI writing or office productivity tool. Chinese-language support, direct access from mainland China, and local payment methods are not disclosed. Users in China may also need to consider NEAR wallet top-ups, access to overseas services, and alternative ecosystem options.