Export

Morphix can export conversations in four formats: Markdown, JSON, PDF, and HTML. Exports include the full conversation history, agent messages, tool outputs, files written, and workflow scorecards — all scoped to the current workspace.

Export Formats

Markdown (.md)

Plain Markdown with conversation structure preserved. Each message is rendered as a block with role, agent name, and content.

# Morphix Conversation — 2026-06-24 14:30

## User
Create a REST API with FastAPI for managing a todo list...

## Assistant (Developer)
Here's the implementation...

## Agent Message (Analista — Subtask 3: Review the API design)
The API structure follows REST conventions...

JSON (.json)

Structured export with full metadata:

{
  "conversation_id": 42,
  "workspace": "main",
  "created_at": "2026-06-24T14:30:00",
  "title": "FastAPI Todo API",
  "messages": [
    {"role": "user", "content": "...", "timestamp": "..."},
    {"role": "assistant", "content": "...", "timestamp": "..."},
    {"role": "agent", "content": "[Analista — Review API]\n...", "timestamp": "..."}
  ],
  "scorecard": {
    "subtasks": 5,
    "completed": 5,
    "tokens": 15234,
    "time": "45.2s"
  },
  "files_written": ["src/main.py", "src/models.py", "tests/test_api.py"]
}

PDF (.pdf)

Formatted PDF with headers, monospace code blocks, and page breaks between major sections. Uses fpdf2 for generation.

HTML (.html)

Standalone HTML page with: - Pygments syntax highlighting for code blocks (Python, shell, JSON, YAML, and others) - Responsive layout with conversation bubbles - Role-based color coding (user = blue, assistant = green, agent = amber, tool = gray) - Print-friendly CSS

Pygments dependency

Pygments is lazy-imported. If not installed, code blocks render without syntax highlighting but the HTML export still works. Install with: pip install pygments

What's Included

Conversation Messages

All messages from the conversation are included:

Role	Description
user	Your original prompts and follow-ups
assistant	Morphix's responses
agent	Internal agent messages (subtask execution, analysis)
tool	Tool execution results (files written, commands run)

Agent Messages

During orchestrated workflows, internal agent-to-agent messages are appended to the export history. These show what each agent did during its subtask:

[Developer — Create SQLAlchemy models for Todo items]
Created models.py with Todo and Tag models using SQLAlchemy 2.0 style...

[Analista — Review the code for issues]
Reviewed 3 files. No security issues found. Suggested adding type hints to query parameters.

Scorecard

Each export includes a workflow scorecard with:

• Subtask count (total, completed, recovered, failed)
• Token usage
• Elapsed time
• Quality rating
• Task type and complexity

Files Written

A list of all files created or modified during the workflow, collected from each agent's files_written output and verified against the actual project directory on disk.

Watermark Stripping

Exports are automatically stripped of internal watermarks and meta-instructions. The stripping process removes:

• Internal system prompts and instructions
• Agent loop guidance text
• Watermark markers in the output

This ensures your exports contain only the conversation and results — clean, professional output suitable for sharing or documentation.

The watermark stripping is controlled by a skip flag: when present in the output, the stripping filter is bypassed to preserve intentional formatting.

Project-Scoped Exports

Exports are project-scoped to prevent cross-project contamination:

• Each export includes only messages from the current conversation
• Workspace-specific conversations never appear in another workspace's exports
• Files written list is filtered to the current project's directory

When no project is selected, files are resolved relative to the workspace memory directory.

How to Export

From the History Tab

1. Open the History tab
2. Select a conversation from the list
3. Click the Descargar (Download) button in the top bar
4. Select your format: md, json, pdf, or html
5. Choose a save location

From the Maestro Tab

1. During or after a workflow, click Descargar in the Maestro top bar (Row 2)
2. Select the format from the dropdown
3. The active conversation is exported

Export Format Selection

In the Maestro tab, the Descargar button shows a format selector:

Option	Format	Best for
md	Markdown	Readable archives, GitHub Gists
json	JSON	Programmatic processing, data analysis
pdf	PDF	Sharing, printing, formal documentation
html	HTML	Browser viewing, syntax-highlighted review

Export Implementation Details

Markdown Export

Builds a Markdown string with headers, code fences, and horizontal rules between major sections. Uses textwrap and shlex for safe formatting.

JSON Export

Serializes the full conversation object with json.dumps(indent=2, ensure_ascii=False). Includes all metadata fields.

PDF Export

Uses fpdf2 (FPDF class) with: - A4 page size, 10mm margins - Helvetica font with monospace fallback for code - Auto page breaks at 20mm from bottom - multi_cell() for wrapping long text

HTML Export

Uses Pygments for syntax highlighting. The HtmlFormatter with noclasses=True embeds styles inline for standalone viewing. Supported languages:

• Python (pygments.lexers.PythonLexer)
• Shell/Bash (pygments.lexers.BashLexer)
• JSON (pygments.lexers.JsonLexer)
• YAML (pygments.lexers.YamlLexer)
• Markdown (pygments.lexers.MarkdownLexer)
• SQL (pygments.lexers.SqlLexer)
• JavaScript, TypeScript, HTML, CSS, and others

If Pygments is not installed, the export falls back to plain <pre><code> blocks without highlighting.