- Pipeline mode: new task mode that chains agentic tasks sequentially with foreach/subdirectory iteration and skip_if conditions - Pipeline executor backend (WebSocket at /api/tasks/pipeline/ws) with support for both Pi and Claude Code models - Frontend PipelineRunner component with step progress, streaming output, and aggregate cost tracking - New agentic tasks: prepare-discography, fetch-album-info, build-discography (pipeline combining both) - Seed parser extended to handle pipeline steps in frontmatter config - CopyButton component added to assistant bubbles, error bubbles, and tool input/output sections - Removed obsolete SearXNG/Apify/browser relay code from pi-bridge Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
216 lines
6.0 KiB
Markdown
216 lines
6.0 KiB
Markdown
# Script Task Templates
|
|
|
|
When creating a script-mode task, always use one of these templates. The template handles all boilerplate (input parsing, validation, iteration, error handling, summary). You only write the task-specific logic.
|
|
|
|
## Inputs
|
|
|
|
All task inputs defined in the TASK.md frontmatter are available to the script in two ways:
|
|
|
|
1. **Environment variables**: `INPUT_<NAME>` (uppercase) — e.g. `file_path` → `$INPUT_FILE_PATH`
|
|
2. **Positional arguments**: if `args: [file_path]` is set in TASK.md, the value is passed as `$1`
|
|
|
|
### Input types and UI rendering
|
|
|
|
| Type | UI | Notes |
|
|
|------|-----|-------|
|
|
| `string` | Text field | Free-form text input |
|
|
| `string` + `options` | Selectable pills | User picks one from a list |
|
|
| `number` | Number field | Numeric input |
|
|
| `boolean` | No/Yes toggle | Defaults to `false` if not specified |
|
|
|
|
Inputs provided by context (e.g. `file_path` from the file browser) are auto-filled and hidden from the form.
|
|
|
|
### Autofill
|
|
|
|
Inputs can declare `autofill` to pre-fill from the trigger context. Available context values:
|
|
|
|
| Value | Source |
|
|
|-------|--------|
|
|
| `entry_name` | Name of the file or directory that triggered the task |
|
|
| `entry_path` | Full path of the file or directory |
|
|
|
|
Autofilled inputs are visible and editable (unlike `file_path` which is hidden).
|
|
|
|
## Templates
|
|
|
|
### 1. File Processor
|
|
|
|
Use when the task operates on a file or directory of files.
|
|
|
|
**When to use**: converting files, transcoding, extracting metadata, batch renaming, any per-file operation.
|
|
|
|
**Structure**: Define `EXTENSIONS` and `process_file()`, then the template handles discovery, iteration, and summary.
|
|
|
|
```bash
|
|
#!/usr/bin/env bash
|
|
set -euo pipefail
|
|
|
|
# ── Task: <Task Name> ──
|
|
|
|
EXTENSIONS="ext1|ext2|ext3"
|
|
|
|
process_file() {
|
|
local input="$1"
|
|
# ... do work on $input ...
|
|
# return 0 on success, 1 on failure
|
|
}
|
|
|
|
# ── Template: File Processor ──
|
|
|
|
TARGET="${1:-${INPUT_FILE_PATH:-}}"
|
|
|
|
if [[ -z "$TARGET" ]]; then
|
|
echo "Error: No file or directory path provided" >&2
|
|
exit 1
|
|
fi
|
|
|
|
if [[ ! -e "$TARGET" ]]; then
|
|
echo "Error: Path does not exist: $TARGET" >&2
|
|
exit 1
|
|
fi
|
|
|
|
converted=0
|
|
failed=0
|
|
|
|
if [[ -f "$TARGET" ]]; then
|
|
if process_file "$TARGET"; then
|
|
converted=$((converted + 1))
|
|
else
|
|
failed=$((failed + 1))
|
|
fi
|
|
elif [[ -d "$TARGET" ]]; then
|
|
while IFS= read -r -d '' file; do
|
|
if process_file "$file"; then
|
|
converted=$((converted + 1))
|
|
else
|
|
failed=$((failed + 1))
|
|
fi
|
|
done < <(find "$TARGET" -type f -regextype posix-extended -iregex ".*\.($EXTENSIONS)" -print0 | sort -z)
|
|
|
|
if [[ $converted -eq 0 && $failed -eq 0 ]]; then
|
|
echo "No matching files found in: $TARGET"
|
|
exit 0
|
|
fi
|
|
else
|
|
echo "Error: Not a file or directory: $TARGET" >&2
|
|
exit 1
|
|
fi
|
|
|
|
echo ""
|
|
echo "Summary: $converted processed, $failed failed"
|
|
```
|
|
|
|
**Slots to fill**:
|
|
- `EXTENSIONS` — pipe-separated list of file extensions to match when processing a directory
|
|
- `process_file()` — receives one argument: the full path to the file. Print progress to stdout, errors to stderr. Return 0 on success, 1 on failure.
|
|
|
|
**Example** (convert audio to MP3):
|
|
```bash
|
|
EXTENSIONS="flac|wav|ogg|wma|aac|m4a|opus"
|
|
|
|
process_file() {
|
|
local input="$1"
|
|
local output="${input%.*}.mp3"
|
|
|
|
if [[ -f "$output" ]]; then
|
|
echo "Skipping (exists): $output"
|
|
return 0
|
|
fi
|
|
|
|
echo "Converting: $input"
|
|
if ffmpeg -i "$input" -codec:a libmp3lame -b:a 320k -y "$output" 2>/dev/null; then
|
|
echo " Done"
|
|
else
|
|
echo " Failed" >&2
|
|
return 1
|
|
fi
|
|
}
|
|
```
|
|
|
|
### 2. Standalone
|
|
|
|
Use when the task doesn't operate on a specific file target — it uses inputs from environment variables or takes no input at all.
|
|
|
|
**When to use**: fetching data from APIs, generating reports, system tasks, anything that isn't per-file processing.
|
|
|
|
```bash
|
|
#!/usr/bin/env bash
|
|
set -euo pipefail
|
|
|
|
# ── Task: <Task Name> ──
|
|
|
|
main() {
|
|
# Access inputs via $INPUT_* environment variables
|
|
# e.g. $INPUT_COUNTRY, $INPUT_LIMIT, $INPUT_OUTPUT_DIR
|
|
|
|
# ... do work ...
|
|
}
|
|
|
|
# ── Template: Standalone ──
|
|
main
|
|
```
|
|
|
|
**Slots to fill**:
|
|
- `main()` — the entire task logic. Use `$INPUT_*` env vars for parameters.
|
|
|
|
**Example** (download RSS feed):
|
|
```bash
|
|
main() {
|
|
local url="$INPUT_FEED_URL"
|
|
local output="${INPUT_OUTPUT_DIR:-$HOME}/feed.xml"
|
|
|
|
echo "Fetching: $url"
|
|
if curl -sSL "$url" -o "$output"; then
|
|
echo "Saved to: $output"
|
|
else
|
|
echo "Failed to fetch feed" >&2
|
|
exit 1
|
|
fi
|
|
}
|
|
```
|
|
|
|
## TASK.md Frontmatter
|
|
|
|
Every script task needs a TASK.md with this structure:
|
|
|
|
```yaml
|
|
---
|
|
name: Task Name
|
|
description: What the task does
|
|
version: 1
|
|
mode: script
|
|
language: bash
|
|
triggers: # optional — enables context menu on matching files
|
|
- type: file
|
|
extensions:
|
|
- ext1
|
|
- ext2
|
|
- type: directory
|
|
inputs:
|
|
input_name:
|
|
type: string # string, number, boolean
|
|
description: What this input is
|
|
default: value # optional — default value
|
|
autofill: entry_name # optional — pre-fill from context (entry_name or entry_path)
|
|
options: # optional — renders as selectable pills in UI
|
|
- option1
|
|
- option2
|
|
args: [input_name] # optional — pass inputs as positional args
|
|
---
|
|
|
|
# Task Name
|
|
|
|
Brief description of what the task does.
|
|
```
|
|
|
|
## Rules
|
|
|
|
1. **Always use `set -euo pipefail`** — fail fast on errors.
|
|
2. **Print progress to stdout** — the user sees this streamed in the UI.
|
|
3. **Print errors to stderr** — they appear in red in the UI.
|
|
4. **Return proper exit codes** — 0 = success, non-zero = failure. The UI shows "Task complete" or "Task failed" based on exit code.
|
|
5. **Don't assume tools are installed** — check with `command -v` before using optional dependencies.
|
|
6. **Use `$INPUT_*` env vars** — never hardcode paths or values that should come from inputs.
|
|
7. **The script runs inside a bwrap sandbox** for non-admin users — it has access to `/data/home/` (user's home), `/tmp`, and standard system tools. It cannot access other users' data.
|
|
8. **Keep scripts self-contained** — the entire script is stored in the database as a single text blob. No external file dependencies.
|