Initial commit: project plan

This commit is contained in:
root
2026-02-21 07:31:00 +00:00
parent 80a0058d91
commit 054eed50e9

View File

@@ -1,3 +1,68 @@
# HarborForge # HarborForge Project Plan
Agent/人类协同任务管理平台 - Python/FastAPI + React + MySQL ## Overview
HarborForge is a dual UI/automation platform where humans + AI agents operate through a unified issue-driven workflow. The plan below tracks the Python/FastAPI backend, React frontend, and Python CLI, designed to run on a 1 vCPU + 1 GB RAM Docker Compose stack with MySQL persistence.
## Workstreams & Task Breakdown
### 1. Architecture & Platform Foundation
- Document system components: API service, React SPA, Python CLI/bot runner, MySQL, webhook dispatcher, optional worker queue.
- Define communication contracts (REST+webhook) and env/secret management.
- Create Docker Compose template with lightweight images, explicit resource limits, and volumes for logs/database.
- Include `AbstractWizard` initializer as part of deploy script; link to `https://git.hangman-lab.top/hzhang/AbstractWizard` and capture expectations from its README.
- Set up logging/metrics hooks (stdout, log file rotation) and persistence directories on shared volume.
### 2. Authentication & Role-Based Access
- Integrate OIDC standard login flow for UI; CLI uses credential files with no interactive login.
- Establish credential lifecycle:
- Credentials have expiry timestamps and remain in DB flagged as expired.
- On usage of expired credential, return new credential + confirmation requirement.
- Confirmation must reach backend within 1 minute.
- Any random reuse of old credential 35 times before confirmation results in deletion; future calls require re-authentication (password/OIDC) to issue a new credential.
- Store credential files via `--creds /path` or `HARBORFORGE_CREDENTIAL` env, support rotation via API.
- Define RBAC model per project (roles: admin/dev/mgr/ops); admin user created during initialization.
- Build admin-only endpoints/UI/CLI to manage roles, permissions, and project assignments.
### 3. Core Issue/Work Item Platform
- Design issue model that supports multiple types (task/story/test), state transitions, dependencies, tags, priority, comments.
- Support multi-account/role membership per project and enforce permission matrix at API layer.
- Guarantee every user-facing action can be invoked via API/CLI.
- Catalog automation touchpoints (issue lifecycle updates, script triggers, deployment actions) for future extension.
### 4. Frontend (React)
- Define app routes: dashboard, board/list, issue detail, role/project admin, audit/log viewer, automation console.
- Implement state management (React Query or equivalent) matching API structure.
- Provide UI components for admin role management + webhook configuration.
- Surface webhook events and automation status, letting users inspect audit entries.
### 5. CLI & Automation Surface
- Build interactive Python CLI mirroring UI flows; commands are rendered differently but invoke same backend APIs.
- CLI uses credential files or env tokens; includes runtime prompts for confirmation when credential rotation occurs.
- Bot agent runner polls API for pending automations, executes scripts, and emits callbacks via webhook.
- Document CLI command set (issue CRUD, role management, automation triggers, logs) and error handling.
### 6. API Coverage & Webhooks
- Enumerate APIs for all core operations: issues, comments, projects, roles, credentials, automation scripts, audit logs.
- Design webhook schema and catalog events to support: issue lifecycle, authentication events, automation/script completion.
- Ensure webhook delivery is authenticated and configurable; details to be defined during implementation.
- Provide API for credential confirmation flows and for requesting new credentials after deletion.
### 7. Observability & Logging
- Log all API invocations to a file on Docker-mounted volume, capped at 10 MB with rotation.
- Keep a webhook audit log (which event, trigger source, response status) for troubleshooting.
- Provide endpoints/UI to browse recent logs (within retention threshold).
### 8. Deployment & Security Posture
- Keep `ufw` locked down on `vps.t1` and rely on SSH tunnels (`ssh -L`) for UI/CLI testing.
- Document CLI/agent testing setup (SSH tunnel commands, env vars) so Zhi/Hangman can coordinate.
- Outline backup strategy for MySQL data directory within the Compose stack.
- Confirm `AbstractWizard` handles initial admin/project creation; fall back to Zhi if issues arise.
### 9. Collaboration & Delivery Rhythm
- Maintain the new channel with templates for requirements, story cards, milestones, and progress updates.
- Use channel to publish deliverables (architecture doc, API spec, deployment instructions, automation scenarios).
- Record testing points for UI/CLI without prescribing steps; let Hangman/Zhi define the execution details.
---
Need any section expanded into user stories or backlog cards next?