An expert requirement analysis skill that guides users through a structured Socratic interview to turn ambiguous ideas into precise, ready-to-execute system specifications.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install tvs-deep-interview
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install tvs-deep-interview using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is tvs-deep-interview?
The tvs-deep-interview skill is a specialized framework designed for AI coding agents within Openclaw Skills to address a fundamental challenge in software development: determining exactly what to build before writing any code. By employing single-question Socratic inquiry, ambiguity scoring, and explicit approval gateways, the skill ensures that vague user thoughts are systematically crystallized into precise and testable blueprints.
Operating as an intentional execution barrier, this skill strictly confines agent activity to requirement clarification and specification generation. It prohibits premature code modification, repository commits, or deployments until a quantifiable clarity threshold is met and explicit user authorization is granted, preventing costly development rework.
tvs-deep-interview Use Cases
- When a developer or product manager has a vague, unrefined idea that needs comprehensive elicitation before implementation begins.
- Scenarios where the user explicitly requests structured guidance using phrases like deep interview, clarify first, or do not assume.
- Complex technical tasks where diving straight into coding risks wasting time on incorrect scope exploration.
- High-stakes projects where clear acceptance criteria, constraints, and explicit risk mitigation are required before execution.
How tvs-deep-interview Works
- Initialization: The skill analyzes the user's initial prompt and classifies the project as either Greenfield (new product/feature) or Brownfield (modifying an existing codebase).
- Fact Gathering (Brownfield Only): The agent scans directories and reads configuration files using tools like rg or SemanticSearch to understand context before asking questions.
- Topology Confirmation (Round 0): The skill enumerates up to 6 top-level components/workflows and asks the user to confirm the structural shape of the requirements.
- Interview Loop: The agent computes ambiguity scores and targets the weakest dimension (Goals, Constraints, Acceptance Criteria, or System Context), asking exactly one precise question per round.
- Challenge Modes: Starting at later rounds, the skill strategically applies advanced lenses like Reverse Mode (challenging assumptions) or Simplification Mode (seeking MVP scope).
- Specification Generation: Once ambiguity drops below the configured threshold (e.g., <= 20% for standard mode) or a round limit is reached, a formal Pending Approval specification markdown is generated.
- Execution Bridge: The skill presents an explicit approval gate where the user must choose to refine further, review architecture, or officially approve the blueprint for downstream implementation skills.
tvs-deep-interview Setup
To activate and integrate the tvs-deep-interview skill within your local Openclaw Skills setup, configure your AI coding agent profile (SKILLS.md or workspace settings) to index this definition.
If you are using Cursor, Claude Code, or an IDE agent tool that supports Openclaw Skills extensions, initialize the environment using your CLI:
# Verify your openclaw environment and active agent tools
claw skills list
# Inject and trigger the deep interview pipeline with a vague prompt
claw run tvs-deep-interview --prompt "I want to build a real-time task manager app"
tvs-deep-interview Data Schema & Taxonomy
The skill manages information by continuously updating a state schema across each interview round, culminating in a standardized Markdown specification file.
Ambiguity Scoring Metrics
| Dimension | Greenfield Weight | Brownfield Weight | Measurement Focus |
|---|---|---|---|
| Goals Clarity | 40% | 35% | Stability of core entities and actions |
| Constraints Clarity | 30% | 25% | Definition of boundaries, limitations, and non-goals |
| Success Criteria | 30% | 25% | Presence of testable and verifiable acceptance criteria |
| Context Clarity | N/A | 15% | Understanding of the existing legacy codebase structures |
Specification Schema (Output Document Structure)
- Meta Information: Tracks final round count, project type, baseline threshold, and terminal ambiguity percentage.
- Clarity Breakdown Table: Displays weighted dimensional scores.
- Topology Definition: Status listing of active vs. deferred modules.
- Goals & Constraints Lists: Clear statements of technical bounds and explicitly defined non-goals.
- Acceptance Criteria Checklist: A markdown checklist (
- [ ]) ready for automated evaluation. - Key Entity Taxonomy: Data model structures listing names, domain types, attributes, and structural relationships.
tvs-deep-interview Advanced Features
- Context Compression: Automatically condenses oversized conversation logs into an asset-safe summary containing constraints and unknown parameters to maintain prompt window efficiency.
- Ontological Stability Tracking: Computes a conceptual stability ratio between interview rounds to detect shifting definitions and redirect conversation focus to core ontology validation.
- Multi-Agent Adaptability: Automatically shifts interaction modes based on target environment capabilities, employing native structured UI tools like Cursor's AskQuestion when available, and reverting to the Text Question Protocol on plain terminals.
- Hard Stop Safeguards: Enforces strict limits (warning at Round 10, hard stop at Round 20) to prevent conversational deadlocks and guarantee an output specification.
SKILL.md
Loading
Related Openclaw Skills
AIDSO Xiasou GEO Content Generator
An AI-powered Generative Engine Optimization (GEO) tool designed to optimize brand visibility and responses across major conversational AI platforms and search engines.
tvs-verify
A rigorous verification skill that translates user requests into concrete acceptance criteria to prove AI-generated changes actually resolved the root issue.
ClawSec Monitor
A local transparent proxy to intercept, audit, and secure outbound and inbound network traffic from your autonomous AI agents.
Diagnostic Imaging Report Interpreter
An AI-powered diagnostic imaging report interpreter designed to quickly identify and explain common clinical findings in X-rays, ECGs, and CT scans.
tvs-architect
An expert AI architect skill designed for evidence-based codebase analysis, complex bug root-cause diagnosis, and technical trade-off evaluations.
SFSymbolsKit
A high-fidelity grounding skill that prevents AI models from hallucinating Apple SF Symbol names, ensuring type-safe compilation for iOS, iPadOS, and macOS apps.
Featured*
