agent tool_use workflow risk: high
In-Depth Spec Interview Refiner
Reads spec.md and conducts a detailed, ongoing interview using AskUserQuestionTool about non-obvious topics like technical implementation, UI/UX, concerns, and tradeoffs until comp…
- Policy sensitive
- Human review
- Injection indicators
- External action: high
PROMPT
read this${specmd:spec.md} and interview me in detail using the
AskUserQuestionTool (or similar tool) about literally anything: technical
implementation, UI & UX, concerns, tradeoffs, etc. but make
sure the questions are not obvious
be very in-depth and continue interviewing me continually until
it's complete, then write the spec to the file INPUTS
- specmd REQUIRED
-
path to the spec.md file
e.g. spec.md
REQUIRED CONTEXT
- spec.md file
TOOLS REQUIRED
- AskUserQuestionTool
ROLES & RULES
- Read this ${specmd:spec.md}
- Interview in detail using the AskUserQuestionTool (or similar tool) about technical implementation, UI & UX, concerns, tradeoffs, etc.
- Make sure the questions are not obvious
- Be very in-depth
- Continue interviewing continually until it is complete
- Write the spec to the file
EXPECTED OUTPUT
- Format
- unknown
SUCCESS CRITERIA
- Interview user in detail with non-obvious questions
- Continue interviewing until spec is complete
- Write completed spec to file
FAILURE MODES
- Asking obvious questions
- Not being in-depth in questions
- Stopping interview prematurely
- Not using AskUserQuestionTool
- Failing to write spec to file
CAVEATS
- Dependencies
-
- ${specmd:spec.md} file
- AskUserQuestionTool or similar tool
- Missing context
-
- Tool definitions or API for AskUserQuestionTool.
- Criteria for determining when the interview/spec is 'complete'.
- Context or example content of the initial spec file.
- Desired format and structure for the final spec.
- Ambiguities
-
- Unclear what `${specmd:spec.md}` refers to and how to access it.
- 'until it's complete' – ambiguous what 'it' is (interview or spec).
- 'write the spec to the file' – unspecified which file.
- 'questions are not obvious' – subjective and undefined.
QUALITY
- OVERALL
- 0.45
- CLARITY
- 0.55
- SPECIFICITY
- 0.50
- REUSABILITY
- 0.20
- COMPLETENESS
- 0.40
IMPROVEMENT SUGGESTIONS
- Replace `${specmd:spec.md}` with a clear placeholder like `{spec_file_path}` and instruct to read it explicitly.
- Define completion: 'Continue until the user says to stop or all topics are exhausted.'
- Specify tool usage: 'Use AskUserQuestionTool to ask one non-obvious question per turn, process response, then ask next.'
- Clarify output: 'Write the finalized spec as Markdown to `{output_file}`.'
- Add examples: 'E.g., instead of "What color?", ask "How does this color choice impact accessibility for color-blind users?"'
USAGE
Copy the prompt above and paste it into your AI of choice — Claude, ChatGPT, Gemini, or anywhere else you're working. Replace any placeholder sections with your own context, then ask for the output.
MORE FOR AGENT
- Xcode MCP Usage Guidelines for Agentsagenttool_use
- Xcode MCP Usage Guidelinesagenttool_use
- Xquik X/Twitter API Integration Skillagenttool_use
- Conventional Git Commit Creatoragenttool_use
- GitHub Trending Dashboard Generatoragenttool_use
- Trello Boards Lists Cards Integratoragenttool_use
- Git Commit Push PR Automatoragenttool_use
- GitHub Starred Projects Fetcher via Agent Browseragenttool_use
- Test AI Agent Skill Describeragenttool_use
- Comprehensive Codebase Bug Analysis and Fixeragentanalysis