agentleFS
Sign inStart free
JetBrainsKnown publisherJetBrains/mcp-steroid/test-integration-agent-launch/CLAUDE.md

mcp-steroid / test-integration-agent-launch

CLAUDE.md82 starsChanged 3 months agoApache-2.0View on GitHub

Cross-OS behaviour tests (Windows and Linux) for the script-only, minimal-dependency way an agent plugin launches things — its stdio MCP server and its hooks — validating how Claude Code resolves each candidate script form per OS. Uses mock probe scripts (no real devrig). Live counterpart to jonnyzzz/mcp-steroid#253. The suite is structurally tied to a real Windows/Linux agent (it downloads + runs the OS Claude build). Per the root CLAUDE.md rule, it is gated at the task level: build.gradle.kts → tasks.test…

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this CLAUDE.md into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    No API key needed — Claude spawns plugin MCP servers + hooks during session init, before the -p turn fails auth.

    line 55

≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
CLAUDE.md

test-integration-agent-launch — Agent Guide

Cross-OS behaviour tests (Windows and Linux) for the script-only, minimal-dependency way an agent plugin launches things — its stdio MCP server and its hooks — validating how Claude Code resolves each candidate script form per OS. Uses mock probe scripts (no real devrig). Live counterpart to jonnyzzz/mcp-steroid#253.

TestVerifies
ClaudeAgentLaunchTestDownloads the host-OS Claude build, loads a mock probe plugin via --plugin-dir, and asserts the per-OS launch outcome of the MCP command + every candidate hook form (exec ×3 + shell ×3).
InstallerScriptTestThe generated installer for the host OS is syntactically valid: install.ps1 parses under Windows PowerShell 5.1; install.sh parses under sh -n. (#254)

OS gating (no runtime skips)

The suite is structurally tied to a real Windows/Linux agent (it downloads + runs the OS Claude build). Per the root CLAUDE.md rule, it is gated at the task level: build.gradle.kts → tasks.test { enabled = isWindows || isLinux }. On macOS the task is skipped; compileTestKotlin still runs. Tests that DO run use OS-conditional assertions (assert the per-OS-correct outcome) — that is not a skip.

Confirmed matrix (cross-validated on TeamCity + GitHub, Windows + Linux)

Probe log: LAUNCHED tag=<tag> via=<cmd|sh|shell|script>.

MCP stdio (script-only, zero deps): extensionless ${CLAUDE_PLUGIN_ROOT}/bin/<stem> resolves to <stem>.cmd via cross-spawn/PATHEXT on Windows (via=cmd) and to the #!/bin/sh <stem> on Unix (via=sh); a bare name resolves on neither (plugin bin/ isn't on the subprocess PATH).

Hooks:

Hook formWindowsLinux
exec-form, extensionless path❌✅ (+x shebang script)
exec-form, .cmd path❌❌
exec-form, cmd /c <.cmd>✅❌ (no cmd)
shell-form, default shell✅ (Git Bash)✅ (sh)
shell-form, "shell":"powershell"✅pwsh-dependent (recorded)
shell-form → #!/bin/sh script (devrig pattern)✅ (Git Bash)✅

Recommended (cross-platform, minimal deps): shell-form pointing at a #!/bin/sh script — runs natively on Linux/macOS and via Git Bash on Windows; exactly how devrig ships its hooks. The single dependency is Git Bash on Windows for sh hooks; MCP has none.

Running

# On Windows or Linux (real behaviour):
./gradlew :test-integration-agent-launch:test
./gradlew ciAgentLaunchTests        # CI aggregator (TeamCity per-OS builds + GH win/ubuntu matrix)

# On macOS: skipped. Verify it still compiles anywhere:
./gradlew :test-integration-agent-launch:compileTestKotlin

No API key needed — Claude spawns plugin MCP servers + hooks during session init, before the -p turn fails auth. TeamCity wiring: ~/Work/mcp-steroid-teamcity → builds/_18_agent_launch_tests.kt.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free