Healthcheck now reads existing Codex Desktop rollout logs and produces a local report of duplicate records, retry chains, and consecutive tool failures. We would like five non-author volunteers to help verify installation and whether those results make sense on their own data.
This is an engineering preview. Codex CLI is not yet independently verified, and Claude Code is not supported. No account, star, or uploaded logs are required.
Try it
With Python 3.9+, Git and pipx installed:
pipx install "git+https://github.com/roy-tong/AgentMeasure#subdirectory=healthcheck"
agentmeasure demo
agentmeasure check --json report.json
The demo is synthetic. Only the second command on your own supported logs is an own-data trial. An installation blocker or an unclear result is useful feedback too.
What would help
- Your OS, Python / AgentMeasure version, and Desktop or CLI runtime.
- Approximate installation time and time to the first own-data report, separately.
- One finding you understood, or the point where you got stuck; what would you do next?
Use the first-run feedback form or reply here. Both are optional. You do not need to post twice.
Keep raw logs, full JSON reports, HTML reports, and snapshots local. They may contain paths and identifiers. If you want to share aggregate counts:
agentmeasure share report.json
agentmeasure share report.json --out summary.md
The first command is a preview. Inspect summary.md yourself before posting it; you can also leave out the summary entirely. Full first-run guide.
If you later change something and run again, a note about whether the snapshot comparison helped a decision is especially useful. A comparison does not by itself prove causal improvement.
Healthcheck now reads existing Codex Desktop rollout logs and produces a local report of duplicate records, retry chains, and consecutive tool failures. We would like five non-author volunteers to help verify installation and whether those results make sense on their own data.
This is an engineering preview. Codex CLI is not yet independently verified, and Claude Code is not supported. No account, star, or uploaded logs are required.
Try it
With Python 3.9+, Git and pipx installed:
pipx install "git+https://github.com/roy-tong/AgentMeasure#subdirectory=healthcheck" agentmeasure demo agentmeasure check --json report.jsonThe demo is synthetic. Only the second command on your own supported logs is an own-data trial. An installation blocker or an unclear result is useful feedback too.
What would help
Use the first-run feedback form or reply here. Both are optional. You do not need to post twice.
Keep raw logs, full JSON reports, HTML reports, and snapshots local. They may contain paths and identifiers. If you want to share aggregate counts:
The first command is a preview. Inspect
summary.mdyourself before posting it; you can also leave out the summary entirely. Full first-run guide.If you later change something and run again, a note about whether the snapshot comparison helped a decision is especially useful. A comparison does not by itself prove causal improvement.