Add “Publish project” and integrate Llama Together AI Coder for project publishing

[meet306]

When we create a project in the app, there is no option to publish the project or to use the “Llama Together AI Coder” to assist with publish-ready changes. Please add a “Publish project” feature and integration with Llama Together AI Coder so teams can publish projects from the UI and optionally use the AI coder to generate or improve publishable assets (README, deployment scripts, metadata).

Current behavior

After creating a project, users can save it as a draft but cannot publish it from the UI.

There is no integration or workflow to request AI assistance (Llama Together AI Coder) to prepare project files for publishing.

Expected behavior

A clear “Publish” action/button on the project page that changes project state from Draft → Published.

Optionally, an AI-assisted workflow using Llama Together AI Coder that:

Suggests or generates a polished README.

Generates basic deployment instructions (e.g., Dockerfile, GitHub Actions, Vercel/Netlify config) based on project type.

Validates minimal publish metadata (name, description, license, tags).

Shows a preview of files the AI would add or modify and requires explicit user approval before making changes.

Proper permission checks so only users with publish rights can publish.

Audit trail / activity log entry recording publish actions and AI-assisted changes.

Steps to reproduce

Create a new project in the app.

Save the project as draft.

Try to find a “Publish” option or an AI assistant workflow for preparing the project for publishing.

Observe that there is no publish action and no Llama Together AI Coder integration.

Proposed implementation suggestions

UI:

Add a “Publish” CTA on the project editor and project overview.

Add a modal “Prepare for publish” that offers two flows: Manual publish or AI-assisted publish.

In the AI flow, show a diff/preview of proposed files and let users accept/reject modifications.

Backend:

New publish endpoint that changes project status and triggers required hooks (CI, public visibility, notifications).

Endpoint to call Llama Together AI Coder API (or internal AI service) to generate file suggestions; store suggestions as pending changes until user approves.

Security & Permissions:

Only owners/maintainers can publish.

Rate-limit or require consent for AI-generated content.

Log AI suggestions and final accepted changes for auditing.

Testing:

Unit + integration tests for publish flow, permission checks, and AI preview/accept flow.

Acceptance criteria

Users can publish a project from the UI (Draft → Published).

Users can optionally invoke Llama Together AI Coder to get suggested publish-related files.

AI suggestions are shown as a preview/diff and require explicit user approval before being applied.

Publish events are recorded in activity logs and respect permissions.

Additional context / notes

If Llama Together AI Coder requires API keys or configuration, add a settings page for admins to configure the service.

Consider a fallback manual checklist for users who do not want to use AI assistance.

Example labels
feature, enhancement, ui, backend, needs-research