Aws Mcp Setup
Configures AWS MCP servers: hosted no-auth docs endpoint or uvx proxy with credentials
Test report
- Verdict
- Tested · Works
- Score
- Tested
- Jul 21, 2026
- Environment
- Claude Code 2.x (agent harness)
- Upstream re-checked
- Aug 10, 2026 · 955f42d
Cloned the repo, installed into a throwaway HOME (SKILL.md landed at ~/.claude/skills/aws-mcp-setup/SKILL.md), parsed the frontmatter with PyYAML (name + 357-char description + allowed-tools; no bundled files to resolve) and spot-checked its three external references: github.com/aws/mcp-proxy-for-aws HTTP 200, pypi mcp-proxy-for-aws HTTP 200 (v1.6.4), knowledge-mcp.global.api.aws HTTP 200. For output I built two real .mcp.json files. BASELINE (my own knowledge, written before applying the body): a stdio server `uvx awslabs.aws-documentation-mcp-server@latest`; I launched it and probed it over stdio — serverInfo awslabs.aws-documentation-mcp-server 1.29.0, tools read_documentation/read_sections/search_table/search_documentation/recommend, and it pulled 42 packages on first run. SKILL artifact (following Step 2's detection gate: `which uvx` found /opt/homebrew/bin/uvx but `aws sts get-caller-identity` failed with "command not found", so the body routes to Option B): {"awsdocs":{"type":"http","url":"https://knowledge-mcp.global.api.aws"}}; a live MCP initialize POST returned serverInfo AWSKnowledgeMCP 1.0.0 and tools/list returned aws___read_documentation, aws___search_documentation, aws___list_regions, aws___get_regional_availability, aws___retrieve_skill — matching the tool names SKILL.md's Step 3 promises. Concrete delta: the skill's config needs zero local runtime and adds list_regions/get_regional_availability/retrieve_skill, while my baseline hard-requires Python+uvx; conversely baseline has read_sections/search_table/recommend that the hosted one lacks, so it is a real but one-sided win, hence 7 not 9. Docs lose a point because the IAM action names (aws-mcp:InvokeMCP/CallReadOnlyTool/CallReadWriteTool) and the `--metadata AWS_REGION=...` proxy flag could not be exercised without an AWS account. Security scan clean: no curl|sh, no base64, no network exfil, no injection text; only note is the plaintext AWS_ACCESS_KEY_ID/SECRET env example, which the body lists after the recommended profile-based option.
Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works
- Installs cleanly 5/5
- Triggers reliably 5/5
- Output vs. baseline 7/10
- Docs & honesty 4/5
What Aws Mcp Setup does
Sets up AWS MCP servers for a Claude Code project, picking between the hosted AWS Knowledge documentation endpoint (no auth, no local runtime) and the full AWS MCP proxy run through uvx with AWS credentials. Includes a detection step that checks for uvx and valid AWS credentials before choosing, the exact mcpServers JSON for each option, config-file precedence (.mcp.json, ~/.claude.json), and a troubleshooting table. Triggers when you ask to configure AWS MCP, add AWS documentation search tools, or debug an aws-mcp/awsdocs server that is not connecting.
How to install Aws Mcp Setup
git clone --depth 1 https://github.com/zxkane/aws-skills.git /tmp/aws-mcp-setup-src
mkdir -p ~/.claude/skills
cp -R /tmp/aws-mcp-setup-src/plugins/aws-common/skills/aws-mcp-setup ~/.claude/skills/aws-mcp-setup
# Skill is documentation-only (single SKILL.md, no scripts or bundled assets).
# Plugin-marketplace alternative: /plugin marketplace add zxkane/aws-skills then /plugin install aws-common@aws-skills
# No deps needed for the no-auth documentation option (remote HTTP MCP endpoint).
# The full AWS MCP option additionally needs: Python 3.10+ with uv/uvx, plus configured AWS credentials.
# Repo also ships aws-cdk-development, aws-sst-development, aws-cost-operations, aws-agentic-ai, aws-serverless-eda under plugins/*/skills/
Skills live in ~/.claude/skills/ (global) or .claude/skills/
(per-project). Restart Claude Code after installing.
Commands — how to trigger Aws Mcp Setup
-
/aws-mcp-setupConfigures AWS MCP servers: hosted no-auth docs endpoint or uvx proxy with credentials
It also activates on plain-language prompts like these:
-
Set up the AWS documentation MCP server -
Configure aws-mcp with my AWS credentials -
Troubleshoot why my AWS MCP connection fails
Frequently asked questions
- Is the Aws Mcp Setup skill free?
- Yes. The skill itself is free from zxkane/aws-skills. SkillProof publishes the install command and an independent test verdict at no cost.
- Does Aws Mcp Setup work with Claude Code?
- We tested it with Claude Code 2.x (agent harness) on Jul 21, 2026. Verdict: Tested · Works. Cloned the repo, installed into a throwaway HOME (SKILL.md landed at ~/.claude/skills/aws-mcp-setup/SKILL.md), parsed the frontmatter with PyYAML (name + 357-char description + allowed-tools; no bundled files to resolve) and spot-checked its three external references: github.com/aws/mcp-proxy-for-aws HTTP 200, pypi mcp-proxy-for-aws HTTP 200 (v1.6.4), knowledge-mcp.global.api.aws HTTP 200. For output I built two real .mcp.json files. BASELINE (my own knowledge, written before applying the body): a stdio server `uvx awslabs.aws-documentation-mcp-server@latest`; I launched it and probed it over stdio — serverInfo awslabs.aws-documentation-mcp-server 1.29.0, tools read_documentation/read_sections/search_table/search_documentation/recommend, and it pulled 42 packages on first run. SKILL artifact (following Step 2's detection gate: `which uvx` found /opt/homebrew/bin/uvx but `aws sts get-caller-identity` failed with "command not found", so the body routes to Option B): {"awsdocs":{"type":"http","url":"https://knowledge-mcp.global.api.aws"}}; a live MCP initialize POST returned serverInfo AWSKnowledgeMCP 1.0.0 and tools/list returned aws___read_documentation, aws___search_documentation, aws___list_regions, aws___get_regional_availability, aws___retrieve_skill — matching the tool names SKILL.md's Step 3 promises. Concrete delta: the skill's config needs zero local runtime and adds list_regions/get_regional_availability/retrieve_skill, while my baseline hard-requires Python+uvx; conversely baseline has read_sections/search_table/recommend that the hosted one lacks, so it is a real but one-sided win, hence 7 not 9. Docs lose a point because the IAM action names (aws-mcp:InvokeMCP/CallReadOnlyTool/CallReadWriteTool) and the `--metadata AWS_REGION=...` proxy flag could not be exercised without an AWS account. Security scan clean: no curl|sh, no base64, no network exfil, no injection text; only note is the plaintext AWS_ACCESS_KEY_ID/SECRET env example, which the body lists after the recommended profile-based option.
- What is the Aws Mcp Setup SkillProof Score?
- 8.4/10 — installs cleanly 5/5, triggers reliably 5/5, output vs. baseline 7/10, docs & honesty 4/5.
- How do I install Aws Mcp Setup?
- Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
- Can I use Aws Mcp Setup with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
- The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.