README.md

# MonkAndMonkey Plugin

Portable OpenAI plugin package for the hosted MonkAndMonkey MCP.

## Capabilities

- tasks: attention list, create, update, reschedule, backlog, complete
- habits: analyze, create, set/toggle/increment daily logs
- journals: analyze, create, update
- goals, streaks, life areas, projects, tags, tracked-time analytics
- configurable actionable productivity brief

All product data continues to flow through the authenticated MonkAndMonkey MCP and FastAPI backend. The plugin contains no credentials.

## Local validation

```bash
python3 -m json.tool plugins/monkandmonkey/plugin.json >/dev/null
python3 -m json.tool plugins/monkandmonkey/mcp.json >/dev/null
git diff --check
```

These checks cover syntax and whitespace. The OpenAI submission portal performs the authoritative package and MCP scans. For local Codex testing, add this repo-owned package to a local marketplace or copy it into the personal plugin directory, install `monkandmonkey`, and start a new task so tools and skills are reloaded.

## ChatGPT developer mode and public submission

Connect `https://monkandmonkey.org/mcp` in ChatGPT developer mode, complete MonkAndMonkey OAuth, and run the test prompts from the manifest in ChatGPT Work. This package is prepared for submission under the individual publisher identity **Oleksii Skurikhin**. The publisher must verify that exact identity in OpenAI Platform before submitting. Public directory publication makes the plugin discoverable in supported ChatGPT and Codex surfaces; workspace administrators can control access for their teams.

The manifest includes five positive and three negative review cases. Run every case on a dedicated fictional review account with suitable sample data before submission. Record the actual ChatGPT Work flow, inspect the entire video, and host a reviewer-accessible recording. Add its URL to `extensions.com.openai.review.demo_recording_url` only after checking the hosted video. Keep reviewer credentials outside the package and enter them only in the portal's private Review details form. Do not use a real user's account.

The dedicated review identity is `chatgpt-review@monkandmonkey.org`. Its password and SHA-256 user-ID fingerprint belong only in ignored `apps/api/.env` as `CHATGPT_REVIEW_PASSWORD` and `CHATGPT_REVIEW_FINGERPRINT`, alongside `CHATGPT_REVIEW_EMAIL`. From the issue worktree, load that environment and run the existing `apps/api/scripts/curated_demo_fixture.py --store chatgpt` dry run before using `--apply`. Check that the fingerprint matches the intended account. The fixture uses the same fictional gallery data as the iOS review workflow: 1 client, 3 projects, 4 goals, 16 tasks, 3 presets, 6 habits, 4 journals, 12 sessions, and 2 goal statistics. Habit log counts vary with the date. Never enter the review password in a committed file, issue, recording, or chat.

Submission still requires the portal's domain challenge, MCP connection and tool scan, automated package checks, and review approval. Upload the plugin ZIP using **With MCP**, then publish the approved version. Do not include any test password, token, private data, or unreviewed video in the ZIP.

Public URLs:

- website: https://monkandmonkey.org
- privacy: https://monkandmonkey.org/privacy
- terms: https://monkandmonkey.org/terms
- support: https://monkandmonkey.org/support