Atlas / MCP servers / silverprotocol / @silverprotocol/typescript-sdk

@silverprotocol/typescript-sdkCAUTION

Name: @silverprotocol/typescript-sdk
Author: silverprotocol

mcp/silverprotocol/typescript-sdk

MCP server by silverprotocol · agent · ai agents · llm · Audited 2026-10-12

⚑ Report / re-scan Source ↗

TypeScript SDK for AgJSON — the normalized wire format for agent framework I/O

Verdict

CAUTION

Grade

Trust score

71 /100

Exposed tools

12 10r · 1w · 1d

Transport

streamable-http

License

MIT

Stars

Overview

From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.

[](https://www.npmjs.com/package/@silverprotocol/core) [](https://www.npmjs.com/package/@silverprotocol/claude-agent-sdk) [](https://www.npmjs.com/package/@silverprotocol/openai-agents) [](https://www.npmjs.com/package/@silverprotocol/google-adk) [](https://www.npmjs.com/package/@silverprotocol/vercel-ai) [](https://www.npmjs.com/package/@silverprotocol/richtext) [](./LICENSE)

The TypeScript SDK for AgJSON — the open, typed transport for normalized agent-framework I/O, unopinionated about storage and rendering. Open source (MIT).

Install

npm install @silverprotocol/core @silverprotocol/claude-agent-sdk
# or openai-agents / google-adk for other frameworks

Requires Node.js 22.12 or later. The packages are ES modules: import them, or require() them from CommonJS, which Node supports for ES modules from 22.12 on.

Packages

`@silverprotocol/core` — the AgJSON types (AgInput, AgEvent,

AgMessage, AgReduceResult), the stateful-per-invoke Normalizer interface (push(native): AgEvent[] / flush(): AgEvent[]), and the normative Reducer and reduce() (stream → object graph).

`@silverprotocol/` — per-framework normalizers

(claude-agent-sdk, openai-agents, google-adk, ...) that translate each framework's native e

Read from source at commit e001ebdf3776OBSERVED · 2026-10-12

Connect

Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.

claude-code

claude mcp add vercel-ai --env ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY} --env GOOGLE_API_KEY=${GOOGLE_API_KEY} -- npx -y @silverprotocol/[email protected]

claude-desktop

{
  "mcpServers": {
    "vercel-ai": {
      "command": "npx",
      "args": [
        "-y",
        "@silverprotocol/[email protected]"
      ],
      "env": {
        "ANTHROPIC_API_KEY": "${ANTHROPIC_API_KEY}",
        "GOOGLE_API_KEY": "${GOOGLE_API_KEY}"
      }
    }
  }
}

Exposed tools (12)

10 read · 1 write · 1 destructive. Blast radius: 1 tool can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.

Tool	Risk	Description
`a.md`	read	The doc.
`calc`	read	adds
`conformance-probe-resource-link.md`	read	A document the find_doc tool links to instead of inlining.
`delete_file`	destructive	Delete a file.
`doc`	read	a doc
`echo`	read	Echo a message.
`lookup`	read	Look up the weather for a city.
`read_mail`	read	Read mail.
`report`	read	the report
`report.pdf`	read	Rendered from the spreadsheet
`save`	write	Save a note.
`search`	read	web search

Trust audit

CAUTIONgrade C · trust 71/100 Install with care. The audit found things worth knowing before you trust its output.

Layer	What it checks	Result
L0	Provenance & inventory	PASS
L1	Static analysis of the code	WARN
L2	Instruction surface (what it tells the agent)	PASS
L3	Class-specific surface	WARN
L4	Behavioural (sandbox)	SKIPPED