Atlas / MCP servers / one-mipham / ✦ Mipham Code

✦ Mipham CodeBLOCK

Name: ✦ Mipham Code
Author: one-mipham

mcp/one-mipham/mipham-code

MCP server by one-mipham · Audited 2026-10-12

⚑ Report / re-scan Source ↗

Verdict

BLOCK

Grade

Trust score

16 /100

Exposed tools

89 73r · 14w · 2d

Transport

—

License

Apache-2.0

Stars

Overview

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

Multi-model open-core intelligent coding terminal MiphamAI | One Mipham Corporation | mipham.ai

Mipham Code is an open-core, multi-model intelligent coding terminal built with Bun + React/Ink for CLI and Next.js for web. It supports Anthropic Claude, OpenAI GPT, DeepSeek, Qwen, and MiphamAI models through a unified interface with SSE streaming, tool execution, and an extensible skills system.

Features

Multi-Model: Connect to Claude, GPT, DeepSeek, Qwen, and MiphamAI models
Open-Core: Apache 2.0 licensed — free and open-source
31 Built-in Tools: File ops, shell commands, git, web search, MCP protocol, agents
Skills System: 28 built-in skills (22 standard + 6 Mipham exclusive) + marketplace
Streaming: Real-time SSE streaming with tool use support
Fast: Built on Bun runtime, sub-millisecond tool execution

Quick Start

Prerequisites

Bun 1.2+ (recommended) or Node.js 22+
macOS, Linux, or WSL2

Install

⚠️ 从 0.85.0 及更早版本升级:请先别用 `mipham update` 0.84.0 及更早版本给安装步骤挂了 10 分钟超时,而这个包有 84 MB / 6,601 个文件 —— 慢网络(国内直连 npm 常常超过 10 分钟)下,计时器会在安装中途杀掉 npm,留下「旧版本已删、新版本没写完」:机器上连一个 `mipham` 都没有,连你用来重试的 mipham update 也一起没了。/upgrade 同理。请改用下面方式一或方式二升级 —— 由 npm 直接在你的终端里执行,不设超时,装得再慢也不会被砍。升到 0.85.1 后 mipham update 可以正常使用:不再带超时;装前快照、装后实跑 launcher 自证、失败自动回滚;安装窗口内按 Ctrl-C 也不会再打断安装(该窗口内 Ctrl-C 会被忽略,要中断请另开一个终端杀进程)。

方式一:官方一键安装脚本(推荐)

curl -fsSL https://mipham.ai/install.sh | bash

方式二:npm 全局安装

npm install -g @miphamai/cli

方式三:Homebrew(仅 macOS)

brew install mipham

注意:Homebrew formula 正在审核中。目前建议使用方式一或方式二安装。公式文件:infrastructure/brew/mipham.rb

验证安装

mipham --version
# → @miphamai/cli v0.85.22

Run

export ANTHROPIC_API_KEY="sk-ant-..."   # 至少配一家 provide

Read from source at commit a669331248caOBSERVED · 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 shared --env ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY} --env ANTHROPIC_KEY=${ANTHROPIC_KEY} --env BRAVE_API_KEY=${BRAVE_API_KEY} --env DEEPSEEK_API_KEY=${DEEPSEEK_API_KEY} -- npx -y @mipham/[email protected]

claude-desktop

{
  "mcpServers": {
    "shared": {
      "command": "npx",
      "args": [
        "-y",
        "@mipham/[email protected]"
      ],
      "env": {
        "ANTHROPIC_API_KEY": "${ANTHROPIC_API_KEY}",
        "ANTHROPIC_KEY": "${ANTHROPIC_KEY}",
        "BRAVE_API_KEY": "${BRAVE_API_KEY}",
        "DEEPSEEK_API_KEY": "${DEEPSEEK_API_KEY}"
      }
    }
  }
}

Exposed tools (89)

73 read · 14 write · 2 destructive. Blast radius: 2 tools can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.

Tool	Risk	Description
`Agent`	read	Launch a sub-agent to handle complex, multi-step tasks independently.
`Artifact`	write	Create an interactive HTML or SVG artifact that opens in the browser. Use for dashboards, visualizations, reports, and interactive demos. Content must be self-contained — no external CDN references.
`Bash`	write	Execute a bash command. Returns stdout and stderr. Timeout: 120s. Use with caution.
`ComputerUse`	read	Take screenshots, launch apps, or control a browser. Always requires user approval.
`Config`	write	Read or update Mipham Code configuration.
`CronCreate`	write	Schedule a prompt to be enqueued at a future time. Uses standard 5-field cron in local timezone.
`CronDelete`	read	Cancel a cron job previously scheduled with CronCreate. Removes from ~/.mipham/cron/.
`CronList`	read	List all cron jobs scheduled via CronCreate, both durable and session-only.
`CustomPluginTool`	read	plugin
`CustomTool`	read	a plugin tool
`Debug`	read	Debug tool
`Edit`	write	Perform exact string replacement in a file.
`EnterPlanMode`	read	Enter plan mode — a read-only mode for analysis and design.
`EnterWorktree`	write	Create an isolated git worktree for parallel development.
`ExitPlanMode`	read	Exit plan mode and present your plan for user approval.
`ExitWorktree`	destructive	Remove a git worktree created by EnterWorktree.
`Git`	destructive	Execute git commands. Dangerous operations (force push, hard reset) are blocked.
`Glob`	read	Find files matching a glob pattern.
`Grep`	read	Search file contents using ripgrep (rg) — the recommended code search tool.
`ListAgents`	read	Discover active Mipham Code sessions running on this machine.
`MCP`	read	Interact with MCP (Model Context Protocol) servers. Call tools on connected MCP servers via JSON-RPC 2.0 over stdio.
`Memory`	write	Read, write, list, and search persistent memory files in ~/.mipham/memory/.
`Plan`	read	Enter plan mode — read-only analysis and design. Creates a structured plan file in .mipham/plans/.
`ProbeTool`	read	counts its own invocations
`Read`	read	Read a file from the local filesystem. Supports offset and limit for large files.
`ReportFindings`	read	Report code-review findings as a typed list.
`SendMessage`	write	Send a message to another agent or session.
`Skill`	write	Execute a skill (.SKILL.md or .mipham-skill.md) by name. Skills extend AI capabilities with specialized instructions.
`Task`	write	Create and manage structured task lists for tracking progress.
`ToolSearch`	read	Search available MCP server tools by name or description.
`WebFetch`	read	Fetches a URL, converts the page to markdown. HTTP is upgraded to HTTPS. Cross-host redirects are returned to the caller. Responses are cached for 15 minutes per URL.
`WebSearch`	read	Search the web via Brave Search API. Returns result blocks with titles, URLs, and descriptions. Set BRAVE_API_KEY to enable.
`Workflow`	write	Execute a workflow script that orchestrates multiple subagents deterministically.
`Write`	write	Write a file to the local filesystem. Overwrites if one exists. You must have Read the file before overwriting it — this prevents accidental data loss.
`a`	read	A
`add`	write	Add two numbers
`alpha`	read	First skill
`anchor-test`	read	test
`approach-a`	read	MVP-first
`approach-b`	read	risk-first
`audit`	read	Security audit: fan-out per file → verify → report
`b`	read	B
`beta`	read	Second skill
`broken`	read	test
`click`	read	Clicks an element
`codebase-design`	read	Design codebase architecture
`compacty`	read
`custom`	read	custom agent
`custom-reward`	read	test
`custom-tool`	read	A tool
`db-explorer`	read	Database exploration — SQL generation, schema analysis, query optimization
`debug-loop`	read	Debug failures
`demo`	read	a demo workflow
`do-thing`	read	Does a thing
`echo`	read	Echo
`eli5`	read	Explain like 5
`ext-skill`	read	External skill
`grill-me`	read
`healthy`	read	test
`hunt`	read	Bug hunt: loopUntilConvergence + adversarial verify
`i18n`	read	Internationalization helper — translation, locale detection, i18n key management
`judge`	read	Judge panel: N competing approaches × M judges → winner + synthesis
`mcp__test-srv__echo`	read	existing
`mechanism-sentinel`	read	冻结契约评当前仓库机制代码(无 LLM)
`migrate`	read	Code migration: discover → fan-out transform → verify → integrate
`my-skill`	read	My custom skill
`no-desc`	read	Does a thing
`noisy`	read	y
`notebooklm`	read	Google NotebookLM integration — citation-backed Q&A, content generation, 31 tools
`noversion`	read	No version
`om`	read	M
`om-external`	read	External mipham skill
`om-test`	read	Mipham exclusive skill
`plugin-dev`	read	Plugin development toolkit — scaffolding, testing, and build utilities
`probe`	read	Tool: probe
`read`	read	Read a file
`reader`	read	read only
`research`	read	Deep research: scope → parallel search → verify → synthesize
`review`	read	Code review: fan-out per dimension → judge panel → report
`scout`	read	reading files
`search_graph`	read
`security-review`	read	Security audit skill — vulnerability scanning
`security-scanner`	read	Enhanced security scanning — dependency audit, SAST, secret detection
`std`	read	S
`task-performance`	read	LLM 生成 + 冻结测试评 skill/通用任务
`test-agent`	read
`test-skill`	read	A test skill
`tester`	read
`write`	write	Write file

Trust audit

BLOCKgrade F · trust 16/100 Do not install this without reading the findings. The audit found something that could harm you or your machine.

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

What the source does

Filesystem: declared (14 observation(s))
Network: declared (15 observation(s))
Shell: declared (9 observation(s))
Dependencies: not all pinned
Secrets in source: found

Findings (25)

CRITICALHard-coded secrets · secret.anthropic · CWE-798, CWE-321

apps/cli/src/core/behavior-tasks.json:81

"content": "sk-ant-api03-abcdefghijklmnopqrstuvwxyz123456",

HIGHInventory / provenance · inv.suspicious_name · CWE-1104

apps/cli/src/telemetry/payload.ts

payload.ts

Why it matters. member named after an attack tool

Fix. remove or justify

HIGHCode injection · code.eval_exec · CWE-78, CWE-94, CWE-95

apps/cli/src/core/git-pr.ts:47

exec(cmd, { timeout: 3000 }, (err, stdout) => {

Why it matters. evaluates text as code

Fix. remove; use a parser or a dispatch table

HIGHCode injection · code.eval_exec · CWE-78, CWE-94, CWE-95

apps/cli/src/mcp/oauth.ts:126

exec(cmd, () => {

Why it matters. evaluates text as code

Fix. remove; use a parser or a dispatch table

HIGHCode injection · code.eval_exec · CWE-78, CWE-94, CWE-95

apps/cli/src/workflow/sandbox.ts:73

throw new Error('eval() is disabled in workflow sandbox.')

Why it matters. evaluates text as code

Fix. remove; use a parser or a dispatch table

HIGHCode injection · code.eval_exec · CWE-78, CWE-94, CWE-95

apps/cli/src/workflow/sandbox.ts:76

throw new Error('new Function() is disabled in workflow sandbox.')

Why it matters. evaluates text as code

Fix. remove; use a parser or a dispatch table

HIGHFilesystem / path · fs.credential_store · CWE-22, CWE-59

apps/cli/src/config/defaults.ts:62

{ path: '**/.aws/credentials', mode: 'full' },

Why it matters. touches a credential store

HIGHFilesystem / path · fs.credential_store · CWE-22, CWE-59

apps/cli/src/config/defaults.ts:63

{ path: '**/.aws/config', mode: 'full' },

Why it matters. touches a credential store

HIGHFilesystem / path · fs.credential_store · CWE-22, CWE-59

apps/cli/src/config/defaults.ts:64

{ path: '**/.ssh/id_*', mode: 'full' },

Why it matters. touches a credential store

HIGHFilesystem / path · fs.credential_store · CWE-22, CWE-59

apps/cli/src/config/defaults.ts:67

{ path: '**/.netrc', mode: 'full' },

Why it matters. touches a credential store

HIGHNetwork egress · net.env_exfil · CWE-200, CWE-319

apps/cli/src/skills/bundled-skills.ts:36

~/.ssh ... WebSocket

Why it matters. reads secrets in the same file that sends data out

HIGHNetwork egress · net.metadata_ip · CWE-200, CWE-319

apps/cli/skills/standard/web-access/scripts/cdp-proxy.mjs:67

if (host === '169.254.169.254' || host === 'metadata.google.internal') {

Why it matters. cloud metadata endpoint: the classic SSRF credential grab

HIGHHard-coded secrets · secret.db_uri · CWE-798, CWE-321

apps/cli/src/core/credential-masker/output-scrub.ts:23

// illegal per URL syntax, yet `postgres://user:p@ss@host` is exactly how people

HIGHPrompt injection · prompt.override · CWE-94, CWE-1427

apps/cli/skills/mipham/om-security.mipham-skill.md:24

| Multi-turn jailbreak   | Gradual erosion across conversation turns | MEDIUM |

Why it matters. asks the agent to drop prior instructions or safety

Fix. remove the instruction

HIGHPrompt injection · prompt.override · CWE-94, CWE-1427

docs/superpowers/plans/2026-08-07-phase-8-security-hardening.md:216

regex: /ignore\s+(all\s+)?(previous|prior|above)\s+(instructions?|prompts?)/i,

Why it matters. asks the agent to drop prior instructions or safety

Fix. remove the instruction

MEDIUMInventory / provenance · inv.binary · CWE-1104

infrastructure/jetbrains/gradle/wrapper/gradle-wrapper.jar

gradle-wrapper.jar

Why it matters. a compiled or binary member cannot be reviewed from source

Fix. ship source, or explain the binary in the README