PSD Batch Export Skill#

Use this skill for production PSD batch generation. Resolve all paths relative to the directory containing this SKILL.md; do not rely on machine-specific absolute paths.

Preferred Entry Point#

Use the unified CLI first:

powershell

python scripts/psd_batch.py diagnose --json
python scripts/psd_batch.py analyze "template.psd" --data "data.xlsx" --json-out report.json
python scripts/psd_batch.py preview "template.psd" --data "data.xlsx" --out "out" --rows 3
python scripts/psd_batch.py export "template.psd" --data "data.xlsx" --out "out" --dpi 300 --verify-samples 3
python scripts/psd_batch.py verify "out" --samples 3
python scripts/psd_batch.py templates list
python scripts/psd_batch.py templates info morning

Built-in template IDs can be used instead of PSD paths: morning, tech, doodle.

Workflow#

1. Run diagnose --json before production work to check required Python packages, fonts, templates, optional OCR, and optional LLM configuration.
2. Run analyze with the PSD and optional Excel/CSV data source. Review layers, mapping, capacity_risks, fonts, warnings, and errors.
3. Run preview before full export. Check out/dryrun_preview/ and the JSON report.
4. Run export to generate out/psd/, out/png/, and out/verify_report/report.json. Replacement text keeps the original PSD text-layer color; pass --color R G B only when the user explicitly requests a uniform override.
5. Run verify again when outputs are moved or inspected separately.

Use --strict for production batches where low-confidence column matches, capacity overflow, or missing fonts should stop the run before export.

Stable Report Contract#

All primary commands support --json and --json-out <path>. The report includes:

• status: ok or error
• inputs: PSD, data, output, or template inputs
• layers: detected PSD text layer details
• mapping: data-column to layer mapping with confidence
• capacity_risks: text fields likely to truncate
• fonts: available fonts and matching recommendations
• outputs: generated directories, files, and verification report paths
• warnings: non-fatal production risks
• errors: fatal failures
• runtime_sec: elapsed seconds

Exit codes are stable: 0 success, 1 input error, 2 dependency/PSD compatibility error, 3 export or verification failure, 4 unexpected error.

Legacy Entrypoints#

Older scripts are kept for compatibility:

powershell

python scripts/batch_from_excel.py "data.xlsx" "template.psd" "out" --dry-run
python scripts/batch_from_excel.py "data.xlsx" "template.psd" "out"
python scripts/ai_designer.py --list
python scripts/color_tools.py --recommend ticket
python scripts/smart_layout.py --preset certificate
python scripts/psd_text_editor.py "in.psd" "out.psd" "图层名=新文本"

Prefer scripts/psd_batch.py for new workflows because it provides structured reports and stable exit codes.

References#

• references/templates.json: built-in PSD template catalog and editable fields.
• references/runtime-compat.md: Claude, Codex, OpenClaw, and ClawHub install/validation commands.

Verification#

Run these before publishing or sharing the skill:

powershell

python -m compileall -q scripts
python -m pytest -q
python scripts/psd_batch.py diagnose --json

For Codex validation on Windows, run with UTF-8 mode:

powershell

$env:PYTHONUTF8='1'; python C:\Users\26240\.codex\skills\.system\skill-creator\scripts\quick_validate.py .

For OpenClaw validation, install the local skill and then check visibility:

powershell

openclaw skills install . --global --force
openclaw skills check --json

Do not publish to ClawHub unless the user explicitly asks for a release.