Iterating With AI And Mcp

Drive Compose HotSwan from an AI agent: edit, hot-reload, screenshot, iterate via MCP

Works with setup

Test report

Verdict
Works with setup
Score
7.6/10
Tested
Jul 21, 2026
Environment
Claude Code 2.x (agent harness)
Upstream re-checked
Aug 10, 2026 · d32cf48

Fetched SKILL.md at hot-reload/iterating-with-ai-and-mcp/SKILL.md via the GitHub tree API; frontmatter parses with name+description and all four cross-linked sibling skills (setting-up-compose-hotswan, understanding-hot-reload-limits, preserving-state-across-reloads, generating-baseline-profiles) returned HTTP 200. No security smells: install-skills.sh uses idempotent symlinks under $HOME, no curl|sh, no blobs. I could not run OUTPUT vs BASELINE because the seven hotswan_* MCP tools require the HotSwan IntelliJ plugin's embedded MCP server plus a live Android app on a device — none available here, so no artifact could be produced. Trigger judgments: 3 HotSwan/AI-iteration phrasings correctly load it, 2 adjacent ones (a LazyColumn recomposition question, a plain HotSwan setup request) correctly do not.

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 5/10
  • Docs & honesty 4/5

What Iterating With AI And Mcp does

Teaches an AI agent to drive the Compose HotSwan IntelliJ plugin's embedded MCP server through an edit-reload-screenshot-evaluate loop, using its seven hotswan_* tools plus snapshot-based rollback and a full-install fallback for schema changes. Triggers when a developer wants AI-driven Jetpack Compose UI iteration, asks whether the agent can screenshot the device, or is wiring a Claude Code/Cursor MCP workflow against a running app.

How to install Iterating With AI And Mcp

git clone --depth 1 https://github.com/skydoves/compose-performance-skills.git /tmp/iterating-with-ai-and-mcp-src
mkdir -p ~/.claude/skills
cp -R /tmp/iterating-with-ai-and-mcp-src/hot-reload/iterating-with-ai-and-mcp ~/.claude/skills/iterating-with-ai-and-mcp
# Runtime deps (NOT installed by the copy above): the skill only works when you also have
#   1. Compose HotSwan IntelliJ/Android Studio plugin installed and active
#      https://plugins.jetbrains.com/plugin/30551-compose-hotswan/
#   2. Its embedded HTTP MCP server registered in your agent (exposes hotswan_get_status,
#      hotswan_reload, hotswan_take_screenshot, hotswan_start_snapshot, hotswan_stop_snapshot,
#      hotswan_select_variant, hotswan_build_and_install)
#   3. A target Android app running on a connected device/emulator, watcher state == WATCHING
# Repo ships scripts/install-skills.sh to symlink ALL 26 skills into ~/.claude/skills at once.

Skills live in ~/.claude/skills/ (global) or .claude/skills/ (per-project). Restart Claude Code after installing.

Commands — how to trigger Iterating With AI And Mcp

  • /iterating-with-ai-and-mcp Drive Compose HotSwan from an AI agent: edit, hot-reload, screenshot, iterate via MCP

It also activates on plain-language prompts like these:

  • Iterate on this Compose screen using HotSwan
  • Hot reload and screenshot this UI change
  • Compare this screenshot against the design intent

Frequently asked questions

Is the Iterating With AI And Mcp skill free?
Yes. The skill itself is free from skydoves/compose-performance-skills. SkillProof publishes the install command and an independent test verdict at no cost.
Does Iterating With AI And Mcp work with Claude Code?
We tested it with Claude Code 2.x (agent harness) on Jul 21, 2026. Verdict: Works with setup. Fetched SKILL.md at hot-reload/iterating-with-ai-and-mcp/SKILL.md via the GitHub tree API; frontmatter parses with name+description and all four cross-linked sibling skills (setting-up-compose-hotswan, understanding-hot-reload-limits, preserving-state-across-reloads, generating-baseline-profiles) returned HTTP 200. No security smells: install-skills.sh uses idempotent symlinks under $HOME, no curl|sh, no blobs. I could not run OUTPUT vs BASELINE because the seven hotswan_* MCP tools require the HotSwan IntelliJ plugin's embedded MCP server plus a live Android app on a device — none available here, so no artifact could be produced. Trigger judgments: 3 HotSwan/AI-iteration phrasings correctly load it, 2 adjacent ones (a LazyColumn recomposition question, a plain HotSwan setup request) correctly do not.
What is the Iterating With AI And Mcp SkillProof Score?
7.6/10 — installs cleanly 5/5, triggers reliably 5/5, output vs. baseline 5/10, docs & honesty 4/5.
How do I install Iterating With AI And Mcp?
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 Iterating With AI And Mcp 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.