Demo
A public issue, from signal to merged fix
This page replays a recorded investigation from our own public plugin repo.
It is not live-mocked data, not a fabricated incident, and not a pipeline event-log dump.
Every citation is a public GitHub URL.
Recorded 2026-08-22T13:03:08Z · ProductionMasterAI/production-master#32, closed by PR #44
Provenance, not a mock
Quick-start named a package that does not exist. The source issue is ProductionMasterAI/production-master#32, closed by PR #44 when the owner merged it. Recorded on our own open-source repository. Issue text, code paths and citations are shown as published; the public sources contained no tokens, secrets, environment values or account IDs. The stages below are reconstructed from that public record; they are not the output of a Production Master pipeline run.
The labelled synthetic sample remains on /product as PM-SAMPLE-001, and its recorded walkthrough plays at the end of this page, labelled synthetic. This replay was made on our own open-source repository, not on a customer incident.
Recorded investigation
Follow the public issue from signal to merge
Each step cites a public GitHub object. There is no fabricated confidence number and no live backend behind this page.
ProductionMasterAI/production-master · 2026-08-17 → 2026-08-22 UTC
Install docs name a package that 404s
A first-time Cursor, Codex, or OpenCode user following docs/user/quick-start.md is told to run npx -y @production-master/client. That package does not exist.
Two user-doc files carry the bad command
docs/user/quick-start.md and docs/user/troubleshooting.md are the cited paths. The issue's verification is grep -rn '@production-master/client' docs/ returning nothing after the fix.
The docs drifted from unpublished adapters
The real adapter packages are @production-master/adapter-{cursor,codex,opencode}. They remain private: true and install via each editor's registration, not npm.
A published adapter package was not simply misnamed
npm view @production-master/adapter-cursor returns 404. Telling users to npx a different scoped name would fail the same way.
The docs instructed a package that was never published
0.1.0 is the first real release, so these are the install instructions a first user follows, and they fail on the first command with a registry 404.
The owner merged the docs fix
PR #44 rewrote both docs to clone → npm run build → editor MCP config at packages/adapter-*/dist/cli.js, then the owner merged it. Nothing in the product flipped the draft to ready.
Exhibits you can open
docs/user/quick-start.md and docs/user/troubleshooting.md instructed Cursor, Codex, and OpenCode users to run npx -y @production-master/client. That package was never published and is not part of the 0.1.0 release. The rejected alternative: A published adapter package was simply misnamed in the docs. npm view @production-master/adapter-cursor returns 404; all three adapters ship private: true, so npx of a different scoped name would fail the same way.
| Id | When | Source | What it shows |
|---|---|---|---|
| E-01 | 2026-08-17T20:14:36Z | GitHub issue #32 | The issue states that docs/user/quick-start.md and docs/user/troubleshooting.md tell users to run npx -y @production-master/client, and that the package does not exist. |
| E-02 | 2026-08-17T20:14:36Z | docs/user/quick-start.md · docs/user/troubleshooting.md | Those two public paths are the install instructions a first user follows. The issue asks grep -rn '@production-master/client' docs/ to return nothing after the fix. |
| E-03 | 2026-08-22T10:54:42Z | PR #44 verification | PR #44 recorded npm view @production-master/adapter-cursor as 404 (private: true is accurate) and npm view production-master as 0.1.1 (pmctl is published). |
| E-04 | 2026-08-22T13:03:08Z | PR #44 merge | The owner merged PR #44 and closed issue #32 at 2026-08-22T13:03:08Z. Two files changed, 50 additions, 13 deletions. grep for the old package name in docs/ is clean. |
A person merged it. Nothing in the product did.
Owner merged PR #44 on 2026-08-22T13:03:08Z. Draft-to-ready is a human GitHub action; there is no product path that flips it. Rewrite both docs to the actual 0.1.0 install path: clone, npm run build, point the editor MCP config at packages/adapter-*/dist/cli.js. PR #44, merged by the owner — never auto-merged.
Product walkthrough
Ninety seconds, one real run
One incident, alert to approval gate. It arrives from Jira and is triaged into a run on its own; specialist agents gather the metrics, logs, deploys and code behind it; the findings become numbered citations; and the one irreversible action waits for a second person to co-sign before anything touches production.
16:9 · 1:30 · no audio — the on-screen captions are part of the recording
Synthetic data · a sample investigation, not a customer's
A payments webhook starts returning 5xx in the EU region after a deploy. Production Master reads the deploy diff, the metric series and the logs, and finds a signature-verification flag that shipped true with the same rollout — every claim in its report linking back to the exhibit it came from. It proposes a Helm rollback, classifies it as irreversible, and holds it until a second approver co-signs.
Would rather read than watch? Step through the same incident in text.
Bring one incident and read it back
This recording is our own public issue.
The closed beta runs the same evidence model against yours.
Self-host via Helm — available in closed beta.