Skip to content

akiiwa-games/AgentCompass / packaging/verify-release

verify-release

Verified

Pre-release behavioral verification for AgentCompass (router + tray) and usage-float. Installs from the built release artifacts (wheel and Inno Setup installer) into clean, foreign-looking machine profiles and Windows Sandbox, proving user-visible behaviors before tagging or syncing to another computer.

Release Verification Skill (verify-release)

Verify AgentCompass (router + tray) and usage-float from their packaged artifacts in a clean, foreign-looking machine profile and Windows Sandbox before cutting a release or syncing to a second host (LITTLESPRING / LS).

Principles (pstack)

  1. Test behavior, not implementation (principle-test-behavior-not-implementation): invoke the installed agentcompass console script, router_gate_hook.py, router_data / router_window, usage_float.py, and AgentCompass.exe against local HTTP fixtures and fake provider CLIs.
  2. Prove it works from packaged artifacts (principle-prove-it-works): build the .whl and AgentCompass-Setup-X.exe from an isolated git copy, install into a fresh venv / Windows Sandbox guest, and verify that site-packages (never the repo checkout) is executing.
  3. Never touch the real user profile: all non-guest harnesses run inside throwaway HOME/USERPROFILE/ROUTER_STATE_DIR directories with sitecustomize.py network interception (sandbox_net.json) and verify real_state_snapshot() is unpolluted afterwards.

Environment & Prerequisites

  • OS: Windows 10/11 with Python 3.11+ (python -X utf8) and Node.js (node) on PATH.
  • Optional for full installer / GUI check: Windows Sandbox (wsb.exe) enabled and Inno Setup installer staged under %TEMP%\agentcompass-tools\innosetup-*.exe.
  • No live credentials or network required: provider HTTP (chatgpt.com, api.anthropic.com, raw.githubusercontent.com, cloudcode-pa.googleapis.com, Magpie gateway) and CLIs (claude, codex, agy) are served by local fixtures inside each harness.

How to Run

1. Fast Foreign-Machine + Zero-Dispatch Suite (every pre-release iteration)

python -X utf8 packaging/verify_release.py

This runs:

  • packaging/foreign_machine_check.py: builds the wheel + isolated usage-float repo, installs into a clean venv, and executes F01–F11 across foreign profiles (--only F10 runs one feature; F09 always runs).
  • packaging/zero_dispatch_check.py: verifies every dispatch surface (D01–D13) from the installed wheel.
  • Pytest behavior & regression suites across setup/, skills/agent-compass/scripts/, tray/, and usage-float/.

2. Full Release Verification (including Windows Sandbox Installer & GUI)

python -X utf8 packaging/verify_release.py --with-windows-sandbox

This additionally boots an offline Windows Sandbox (packaging/windows_sandbox_check.py --connect), installs the wheel and builds/installs AgentCompass-Setup-<version>.exe via Inno Setup, drives the tray GUI and router_window, and verifies clean uninstall.

3. Second-Host (LITTLESPRING / LS) Post-Release Acceptance

After pushing the release tag and waiting for .github/workflows/release.yml to publish the release assets:

  1. Upgrade the wheel on LS (pip install --upgrade <wheel or git tag>) and update usage-float.
  2. Run agentcompass data --refresh, agentcompass doctor, agentcompass providers, agentcompass usage, and agentcompass plan --demo --dry-run on LS and verify zero errors, no upgrade-check items, a Reference data: line naming the release's published set (schema as in references/public-data.json), and fresh quota readings.

Feature Inventory

Tracked in features.json. Update both features.json and packaging/foreign_machine_check.py whenever a user-visible routing, quota, gate, setup, tray, or usage-float behavior changes (maintain-verification-skill).

reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 468kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 319kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 319kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 319kUpdated Claude CodeCodexVerifiedInstall