Amp

Amp AGENTS.md, settings.json and MCP

How does Amp load AGENTS.md files, where is settings.json, and how do I add MCP servers?

Amp always loads AGENTS.md from the working directory and its parents, plus your personal and system-wide files, and pulls in subtree AGENTS.md files when it reads files there. User settings live in ~/.config/amp/settings.json, workspace settings in the nearest .amp/settings.json, and every key uses the amp. prefix. Add MCP servers with amp mcp add or the amp.mcpServers setting.

How Amp finds AGENTS.md

  • AGENTS.md in the current working directory (or editor workspace roots) and parent directories up to $HOME is always included.
  • AGENTS.md files in subtrees are included when the agent reads a file in that subtree.
  • $HOME/.config/amp/AGENTS.md and $HOME/.config/AGENTS.md are always included if they exist.
  • System-wide files are always included: /etc/ampcode/AGENTS.md on Linux, /Library/Application Support/ampcode/AGENTS.md on macOS, %ProgramData%\ampcode\AGENTS.md on Windows.
  • If a directory has no AGENTS.md but has AGENT.md or CLAUDE.md, that file is used instead.

Set AMP_IGNORE_GUIDANCE_FILES to a colon-separated list of globs to skip matching files, or to * to skip all of them.

@-mentions and glob-scoped guidance

Mention other files with @ inside an AGENTS.md to include them. Relative paths resolve from the file that contains the mention, absolute paths and @~/ paths work, glob patterns such as @doc/*.md are supported, and mentions inside code blocks are ignored.

A mentioned file can carry globs in YAML front matter. It is then included only after Amp has read a file matching one of the globs. Globs get an implicit **/ prefix unless they start with ./ or ../.

docs/typescript-conventions.md

---
globs:
  - '**/*.ts'
  - '**/*.tsx'
---
Follow these TypeScript conventions:
- Never use the any type

settings.json and key settings

User settings are at ~/.config/amp/settings.json (or settings.jsonc) on macOS and Linux, and %USERPROFILE%\.config\amp\settings.json on Windows. Workspace settings are the nearest .amp/settings.json, searched upward to the repository root, and they override user settings. Run amp config edit to open user settings, or add --workspace for workspace settings.

  • amp.mcpServers: local MCP server definitions.
  • amp.mcpPermissions: ordered allow or reject rules that match MCP servers by command and args or by url. The first match wins and unmatched servers are allowed.
  • amp.tools.disable: tool names to disable, with builtin: prefixes and * globs supported.
  • amp.skills.path: extra skill directories.
  • amp.showCosts, amp.updates.mode and amp.git.commit.coauthor.enabled for costs, updates and commit trailers.

Amp does not ask for approval before running tools. To control tool use, the docs point you to a custom plugin, which workspace admins can distribute as a global workspace plugin.

MCP servers, modes and execute mode

Add a local or URL-based MCP server from the CLI, or define it in amp.mcpServers using command, args and env for local servers or url and headers for remote ones. Use ${VAR_NAME} for environment variables. The docs recommend bundling MCP servers in skills where possible so tools load only when needed.

Add MCP servers

amp mcp add context7 -- npx -y @upstash/context7-mcp
amp mcp add linear https://mcp.linear.app/sse

Modes are set on The Dial: low, medium (the default), high and ultra. A thread keeps the mode chosen before its first message. In the CLI, press Ctrl+S before your first message to change it.

Execute mode runs one prompt, prints the final message and exits. Start it with -x or --execute, by piping input, or by redirecting stdout. For CI, set AMP_API_KEY to an access token from Settings.

Execute mode

amp -x "what package manager is used here?"
cat README.md | amp -x "summarize this file"
amp --mcp-config '{"everything": {"command": "npx", "args": ["-y", "@modelcontextprotocol/server-everything"]}}' -x "What tools are available to you?"

Sources

More on Amp

Other guides

Cursor: allow all terminal commandsClaude Code: allow commands without promptsCodex CLI: approval and sandbox modesGemini CLI: YOLO mode, auto_edit and allowing specific shell commandsCursor Privacy Mode: telemetry, training and data retentionClaude Code telemetry and data retention: what is sent and how to turn it offGemini CLI: turn off usage statistics and telemetryCLAUDE.md: where it goes, how it loads, and how it works with AGENTS.mdClaude Code hooks: format on save, block risky edits, get notifiedClaude Code subagents: create one, limit its tools, and call itAdd MCP servers to Cursor with mcp.json: Keep files out of Cursor with .cursorignore: Configure Codex with config.toml: Configure Gemini CLI with settings.json: Add MCP servers to Claude Code (claude mcp add, .mcp.json, scopes): Claude Code custom slash commands and skills (SKILL.md): GitHub Copilot custom instructions: Adding MCP servers to GitHub Copilot: Claude Code settings.json: file locations, precedence, and key settingsClaude Code plugins and marketplaces: install, create, and shareGemini CLI extensions and custom commands: Configuring OpenCode with opencode.json: Cline Rules: workspace, global, and conditional rulesKiro steering files: .kiro/steering, inclusion modes, and AGENTS.mdZed agent: instruction files, MCP servers, tool permissions, and ACP agentsCursor CLI: install, headless mode, permissions and CIAider: CONVENTIONS.md, .aider.conf.yml and API keysJunie guidelines, MCP and the Action Allowlist: Configure Goose extensions, hints, recipes and permissions: Customize OpenHands with skills, AGENTS.md, setup.sh and MCP: Configure Factory Droid CLI: AGENTS.md, settings.json, autonomy, custom droids and MCPConfigure Qwen Code: settings.json, providers, QWEN.md, MCP and approval modes

Get the weekly agent stack update

New official MCP servers, spec changes and harness releases, checked against the source. One email a week, no fluff.

Reviewed Oct 6, 2026. Settings change often; the linked vendor docs are the source of truth.