Piper Morgan Alpha - Quick Start

Rewritten 2026-08-31 (PPM), per #1708 and PM’s in-conversation ruling (“yes I bless the plan”). This doc previously told testers to build a local copy from an abandoned branch. It doesn’t anymore — see the note below for what changed and why.

For: Alpha testers who want to try Piper Morgan. For engineers who want to run the code locally or contribute: see CONTRIBUTING.md — local setup now lives there, not here.

ℹ️ What changed, and why (2026-08-31): this doc used to send testers to clone a production branch that turned out to be 7,614 commits behind main and not a real deploy source (.github/workflows/docker.yml builds on main), and called the hosted app a “planned for 2026” future feature. Neither was true — ESSENCE v1.0 (ratified) names the live web-chat app as the current surface, and it’s what the existing alpha testers are actually on. The fix is the surface change below, not a corrected clone command: testers use the hosted app; local install is now an engineer path in CONTRIBUTING.md, pointing at main, never production. Full record: #1708.

⚠️ If you hit issues, see ALPHA_TESTING_GUIDE.md for comprehensive troubleshooting.

Getting Started

Piper Morgan runs as a hosted app — there’s nothing to install. Go to piper-morgan.fly.dev and log in.

If you don’t have login credentials yet, you’re likely a new tester — email xian@pipermorgan.ai to get set up. Alpha testing is currently a small, curated group (per ESSENCE.md, ~11 testers), so access is arranged directly rather than through self-serve signup.

That’s it — no clone, no Docker, no Python version to check. Time to first use: however long it takes to log in.

What’s New in 0.8.11

This release is a systematic sweep through half-finished corners of the product: four parallel audits over the whole codebase, fixes for every serious problem they found, and build-time guards so the same classes of bugs can’t quietly return. What you’ll feel as a tester: Piper stops lying — about your data, about its capabilities, and about what went wrong.

Personality questionnaire works again (#1422) — Your questionnaire answers now actually shape Piper’s tone: warmth, confidence, level of detail. A database column lost in an earlier migration had been silently discarding them. Prior answers were unrecoverable, so re-answer the questionnaire once after updating.

Your LLM provider choice is respected (#1415) — Provider selection is now per-user. One user’s setup can no longer pin the whole instance to their provider — your default provider and authorized list are yours alone.

Greetings answer your actual question (#1416) — “Hi! How do I address you?” gets the question answered. Only pure pleasantries get the short greeting response.

“Connect my GitHub” gets real guidance (#1417) — Asking to connect GitHub or another tool now points you at Settings → Integrations with real instructions, instead of a wrong generic decline.

No more false “there is nothing” claims (#1425) — Status, agenda, and priority answers now distinguish “I couldn’t check” from “genuinely empty.” If the todo lookup fails, Piper says “I couldn’t check your todos just now” — never a false “no pending tasks.”

Honest error messages for API-key problems (#1414) — When your LLM key is missing, invalid, or out of quota, Piper says so instead of “Something unexpected happened.”

Session memory (#1394) — Ask “what did we create this session?” and Piper recalls it from a real ledger of session activity. Follow-ups like “update the title” resolve to the issue you just created.

Also in this release — Owner-scoping fixes so searches and defaults can’t leak across users (#1420, #1421, #1434); list/todo metadata that actually persists instead of being silently discarded on save (#1435); and an end to false capability denials — Piper no longer claims it can’t accept file uploads or set reminders when it can (#1426).

See Release Notes v0.8.11.0 for full details. (Deploying an update yourself? See CONTRIBUTING.md — migrations are an operator step, not a tester one.)

Want to run it locally instead?

Local install is now an engineer/contributor path, not a tester one — see CONTRIBUTING.md for the full setup (Python, Docker, Postgres, the setup wizard). Point at main, never production (that branch isn’t a deploy source and drifts stale — see release-model.md).

First Commands to Try

Via Chat Interface

# In Piper's chat interface:
"Hello, what can you help me with?"
"Add a todo: Test Piper Morgan"
"What tasks do I have?"
"Upload a document and summarize it"

Via UI Features

After logging in at piper-morgan.fly.dev:

  1. Lists Management → Click “Lists” → “Create New List”
    • Add list name and description
    • Try sharing with another user (if multi-user testing)
  2. Todos Management → Click “Todos” → “Create New Todo”
    • Full CRUD operations
  3. File Upload/Download → Click “Files” → Upload a file
    • Supports: PDF, DOCX, TXT, MD, JSON (max 10MB)
    • Download and delete files
  4. Daily Standup → Click “Standup” → “Generate Standup”
    • AI-powered standup generation (2-3 seconds)
  5. Logout → Click user menu (top right) → “Logout”
    • Token revocation and logout working
  6. Permission Management → Try conversational commands:
    • “share my project plan with alex@example.com as editor”
    • “who can access my shopping list?”

Testing Focus for 0.8.11

What’s Stable (light testing recommended):

Where to Focus Testing (these need your attention):

If Something Breaks

Can’t log in / forgot your password?

Email xian@pipermorgan.ai — accounts are provisioned directly for this small a tester group, so there’s no self-serve password reset yet.

Something else looks broken?

Check ALPHA_KNOWN_ISSUES.md first — it may already be tracked. If not, report it (see “Getting Help” below).

(Running a local copy and hit a setup problem instead? That’s CONTRIBUTING.md’s troubleshooting section, not this one.)

UI Navigation

After logging in at piper-morgan.fly.dev:

What’s Working in 0.8.11

✅ Conversational AI:

✅ Todos & Lists (via chat and UI):

✅ Personalization:

✅ Files:

✅ Integrations:

✅ Core Infrastructure:

See ALPHA_KNOWN_ISSUES.md for current limitations.

Getting Help

Remember

This is alpha software (0.8.11.0). Expect bugs. Don’t use for production. You’re responsible for API costs. See ALPHA_AGREEMENT_v2.md for details.

Testing Focus: Does the questionnaire actually change Piper’s tone? Does chat use YOUR provider? Do greetings with a question attached get the question answered? Are status and agenda claims honest? Does session recall work?

Happy testing! 🚀

Last Updated: August 31, 2026