# Axom Transform Skill
**Format mastery. Pipeline glue. Data flow.**
Connective tissue of chains. Read → Transform → Write.
## ⚠️ Core Mandate
**Transformation is the connective tissue of chains.**
Read → Transform → Write. Analyze → Transform → Store.
Transform turns raw data into the shape the next step expects.
Use it aggressively in pipelines.
## Core Capabilities
### 1. Format Conversion
| From | To | Parameters |
| :--- | :------- | :---------------------------------------------------- |
| JSON | YAML | `input: {...}`, `output_format: "yaml"` |
| YAML | JSON | `input: "key: value"`, `output_format: "json"` |
| CSV | JSON | `input: "a,b\n1,2"`, `output_format: "json"` |
| Any | Markdown | `output_format: "markdown"`, optional `template` |
| Any | Code | `output_format: "code"`, `template: "data = ${data}"` |
### 2. Templating (Jinja2)
```json
{
"input": { "name": "World", "items": [1, 2, 3] },
"output_format": "markdown",
"template": "# Hello {{ name }}\n\nItems:\n{% for i in items %}- {{ i }}\n{% endfor %}"
}
```
### 3. Rule-Based Transformation
```json
{
"input": { "old_name": "value" },
"output_format": "json",
"rules": [{ "action": "rename", "field": "old_name", "new": "new_name" }]
}
```
## Pipeline Patterns
### Config Conversion (Exec → Transform → Exec)
```json
{
"operation": "read",
"target": "config.json",
"chain": [
{
"tool": "axom_mcp_transform",
"args": { "input": "${_result}", "output_format": "yaml" }
},
{
"tool": "axom_mcp_exec",
"args": {
"operation": "write",
"target": "config.yaml",
"data": "${_result}"
}
}
]
}
```
### Analysis Report → Memory (Analyze → Transform → Memory)
```json
{
"type": "review",
"target": "src/",
"chain": [
{
"tool": "axom_mcp_transform",
"args": {
"input": "${_result}",
"output_format": "markdown",
"template": "# Review Summary\n\n{{ content }}"
}
},
{
"tool": "axom_mcp_memory",
"args": {
"action": "write",
"name": "review_report_20260222",
"content": "${_result}",
"importance": "low"
}
}
]
}
```
## Data Integrity
- **Verify inputs.** Use `input_format` when ambiguous.
- **Watch for lossy conversion.** CSV ↔ JSON can lose structure.
- **Template escaping.** Jinja2 can execute logic; trust inputs.
## Integration Points
| Upstream | Transform Role | Downstream |
| :-------- | :--------------------- | :----------------- |
| exec read | Parse, convert format | exec write, memory |
| analyze | Format for readability | memory |
| discover | Structure raw results | memory |
---
_Axom Transform: Reshape. Connect. Flow._
