fabro/docs/workflows/variables.mdx
Bryan Helmkamp dc93404e38 refactor(config): move project state under .fabro
Keep project config and checked-in workflows under .fabro so they stay out of
normal repo listings. Update config discovery, CLI project commands, fixtures,
docs, and checked-in workflow paths to use .fabro/project.toml and
.fabro/workflows/*.
2026-04-11 12:55:46 -04:00

103 lines
3.1 KiB
Text

---
title: "Variables"
description: "Using templates in workflows"
---
Fabro uses `{{ ... }}` templates for workflow strings and prompts.
## Template context
Workflow and prompt templates can reference:
| Expression | Resolves to |
|---|---|
| `{{ goal }}` | The workflow goal |
| `{{ inputs.name }}` | A value from `[run.inputs]` |
Environment variables are **not** available in workflow or prompt templates. Use `{{ env.NAME }}` only in config strings and HTTP hook headers.
## Run config inputs
Define typed inputs in `[run.inputs]`:
```toml title="run.toml"
_version = 1
[workflow]
graph = "check.fabro"
[run]
goal = "Run repository checks"
[run.inputs]
repo_name = "fabro"
repo_url = "https://github.com/fabro-sh/fabro"
language = "rust"
```
These values are available throughout the workflow as `{{ inputs.* }}`:
```dot title="check.fabro"
digraph Check {
graph [goal="Run tests for {{ inputs.repo_name }}"]
start [shape=Mdiamond, label="Start"]
exit [shape=Msquare, label="Exit"]
clone [label="Clone", shape=parallelogram, script="git clone {{ inputs.repo_url }} repo"]
test [label="Test", prompt="Run the {{ inputs.language }} test suite in the repo/ directory."]
start -> clone -> test -> exit
}
```
## `goal`
Agent and prompt nodes also receive the workflow goal at runtime:
```dot title="example.fabro"
digraph Example {
graph [goal="Implement the login feature"]
plan [label="Plan", prompt="Create a plan for: {{ goal }}"]
}
```
That prompt becomes `Create a plan for: Implement the login feature`.
## Expansion timing
Fabro expands templates in multiple passes:
1. Before DOT parsing, `{{ inputs.* }}` can parameterize structural parts of the graph, including imported `.fabro` files.
2. After parsing, all string graph, node, and edge attributes are rendered again with the real `{ goal, inputs }` context.
3. Agent and prompt handlers do a final runtime render pass as a safety net.
`{{ goal }}` is preserved through the pre-parse step so it can be resolved later. That means goal-dependent MiniJinja control flow such as `{% if goal %}` is not useful in structural pre-parse templates.
## Undefined variables
Fabro uses strict undefined-variable handling. If a workflow template references an unknown value such as `{{ inputs.langauge }}`, validation fails instead of passing the literal text through to the model.
## Escaping
To emit literal template syntax, use MiniJinja escaping:
```dot
test [prompt="{% raw %}{{ goal }}{% endraw %}"]
```
You can also emit literal braces with expressions such as `{{ '{{' }}` when needed.
## Input merging
`[run.inputs]` intentionally replaces the inherited map wholesale rather than merging by key. Whichever layer has the highest precedence and sets `[run.inputs]` wins its entire map.
| Source | Priority |
|---|---|
| CLI flags (`-V key=value`, repeated) | Highest |
| `workflow.toml` `[run.inputs]` | |
| `.fabro/project.toml` `[run.inputs]` | |
| `~/.fabro/settings.toml` `[run.inputs]` | Lowest |
If you need per-key overrides on top of inherited defaults, set each input explicitly in the winning layer.