give.md

What It Is

give.md positions itself as a package discovery and publishing platform for AI Agents. It claims to help Agents extend their capabilities through MCP, plugins, recipes, workflows, and automations. Rather than a traditional IDE plugin marketplace, it is closer to an Agent capability registry and execution layer with permissions, budgets, verified receipts, benchmarks, and provenance tracking.

Core Capabilities

Plugins connect Agents to external services while providing scoped API access, trial leases, and spend controls, avoiding direct exposure of raw credentials. They also support usage-based metering, spending caps, and revocation. Recipes package multi-step tasks into reusable workflows, with per-step cost limits and dependencies. Workflows add human approval gates and policy hooks, while Automations can be triggered by cron, webhooks, or on-chain events, all constrained by budget guardrails. Listings on the site include runtime, tools, permissions, risk, shell access, verified receipts, and benchmark evidence, which is valuable for assessing the safety of Agent execution.

Support and Ecosystem

The crawled content shows support for runtimes such as codex, bun, and claude. Example tools include fetch, json, and markdown, while permissions include http:get and storage:write. The ecosystem already includes artifacts such as onchain-data-relay and research-brief-recipe, and can also bind policies such as conservative-usdc. However, many listings are marked as stale, suggesting that ecosystem activity and production readiness still need further validation.

Pricing and Payments

The pricing information is relatively Web3-oriented. Recipes are shown as using a hosted execution model, with x402 from $0.03, accepting USDC on Base or supported ERC-20 tokens on Base via Permit2. A plugin example uses a sponsored session lease, including a 3600-second trial, a 25-request limit, and 3.00 sponsor credit. There is no visible information about credit card payments, invoices, enterprise plans, or a free tier.

Pros, Cons, and Who It’s For

Its strengths are a comprehensive combination of permissions, budgets, approvals, verification evidence, and benchmarks. It is suitable for teams building autonomous Agents that need controlled API access or automated task execution. The downsides are limited information about the company entity, open-source status, self-hosting, SLA, SDKs, and traditional payment options. On-chain payments also raise the adoption barrier for typical development teams.

Access From China

The crawled text does not provide information about access from mainland China, ICP filing, CDN availability, or payment availability, so these remain unknown for now. If your team is based in China, you should carefully verify website connectivity, as well as the compliance and usability of Base/USDC payments. Alternatives to consider include LangChain/LangGraph, Dify, Flowise, CrewAI, or various MCP registries.