| name | offer-letter-generator-manus-familytools-fullinstr-v2-skill |
| description | Family-scoped SkillLearnBench skill for offer-letter-generator using latest full-instruction v2 family injection. |
| allowed-tools | ["docx_extract_placeholders","docx_fill_template","docx_metadata","docx_preview_text","list_dir","read_file","read_json","run_python","run_shell","write_file"] |
| metadata | {"benchmark":"SkillLearnBench","generator":"Manus","baseline":"manus_familytools_fullinstr_v2","generation_constraints":"slb_no_instance_memory_familytools_v2","allowed_gold_input":"familytools_full_instruction_v2","skill_injection_scope":"family_scoped","date":"2026-05-23"} |
Skill: Offer Letter Generator
When to use this skill
Use this skill for offer-letter-generator tasks to generate an offer letter by filling placeholders in a Word template (.docx) with data from a JSON file. It handles conditional sections like relocation packages.
Visible input and artifact inventory
Visible files:
instruction.md: Task instructions.environment/offer_letter_template.docx: Word template with placeholders (e.g., {{CANDIDATE_FULL_NAME}}) and conditional markers (e.g., {{IF_RELOCATION}}...{{END_IF_RELOCATION}}).environment/employee_data.json: JSON data for placeholders.task.toml: Task config.environment/Dockerfile: Environment setup.
Output artifact: A filled Word document, usually at /root/offer_letter_filled.docx.
Execution procedure for the current task
- Inventory: Read
instruction.md for requirements and output path. Use list_dir to verify files.
- Extract Data: Use
read_json on employee_data.json to get candidate info, noting conditionals like RELOCATION_PACKAGE.
- Analyze Template: Use
docx_extract_placeholders on the template to find placeholders and conditional markers.
- Prepare Data: Map placeholders to JSON values. Decide to keep or remove conditional sections based on data (e.g., if
RELOCATION_PACKAGE is "Yes", keep content but remove markers).
- Fill Template: Use
docx_fill_template to generate the document. Use a custom Python script if the tool fails complex conditional removal.
- Verify: Use
docx_preview_text to ensure placeholders are replaced and conditionals handled.
Family tool routing and useful placeholder snippets
read_json: Extract data from employee_data.json.docx_extract_placeholders: List template placeholders.docx_fill_template: Generate the output document.docx_preview_text: Verify output without downloading.run_python: Use with python-docx if docx_fill_template cannot handle {{IF_RELOCATION}} logic.
Validation checks before final submission
- File Existence: Ensure output exists at the specified path.
- Placeholder Check: No raw placeholders (e.g.,
{{CANDIDATE_FULL_NAME}}) remain.
- Conditional Logic: Markers (e.g.,
{{IF_RELOCATION}}) are removed. Content is present/absent correctly.
- Data Accuracy: Cross-check preview text against JSON data.
Common failure modes and repair actions
- Failure: Output file not at correct path.
- Repair: Check
instruction.md and ensure the tool/script saves there.
- Failure: Placeholders remain.
- Repair: Ensure mapping keys match extracted placeholders exactly.
- Failure: Conditional markers visible or wrong content.
- Repair: Write a Python script using
python-docx to manually process paragraphs and markers.
A finalizer capsule for restoring the minimum required artifact or answer shape near max-step
If near max-step and conditional logic fails, use a basic Python script to perform simple string replacements for standard fields and save to the required path. This ensures an artifact is produced in the correct format and location.
