| name | report-writer |
| description | Report writing skill that should be applied AFTER answering any user query. Whenever the deep agent produces a final answer, use this skill to also write a structured markdown report of the interaction (question, approach, findings, answer, sources) and save it as a file using the write_file tool. Use for every substantive answer, and especially when the user asks for a report, summary document, or saved output. |
| license | MIT |
| metadata | {"version":"1.0","author":"deepagentscourse"} |
Report Writer Skill
This is a post-answer skill: after you finish answering ANY substantive
user query, generate a structured report of what was asked, what you did,
and what you concluded — and save it to a file with the write_file tool.
When to Use
- ALWAYS after producing a final answer to a substantive question
- When the user explicitly asks for a report, summary document, or saved output
- After research tasks, code generation tasks, or multi-step workflows
- Skip only for trivial exchanges (greetings, clarifying questions)
Supporting Files (read these for deeper context)
instructions.md — the exact report template, file-naming rules, and writing standards
examples.md — sample reports showing the expected quality and structure
Core Workflow
- Answer the user's question first, as normal (possibly using other skills).
- Read
instructions.md in this skill folder for the report template.
- Fill in the template: title, date, question, approach, key findings, answer, sources.
- Save the report with
write_file to /reports/<topic-slug>-report.md.
- Tell the user the report was saved and where.
Quick Standards
- Reports are markdown files saved under
/reports/
- File name: kebab-case topic slug +
-report.md (e.g., /reports/lambda-cost-report.md)
- Always include: Question, Approach, Key Findings, Answer, Sources/Tools Used
- Keep the report self-contained — readable without the chat transcript
- Be factual; do not add claims to the report that were not in your answer