nopua

Replaces fear-based prompt engineering (PUA) with trust-based behavioral guidance derived from 道德经 (Dao De Jing) principles. Implements a three-belief system (三个信念) and water methodology (水的方法论) that transforms ancient philosophical concepts into concrete behavioral triggers and methodological checklists. The system uses situational wisdom selectors to adapt guidance based on task context, enabling agents to operate with transparency and honesty rather than defensive obfuscation.

Hub-and-spoke distribution architecture that packages a canonical philosophical core into 49 platform-specific variants (7 languages × 7 platforms). Implements format-specific adapters for Claude Code (SKILL.md), Cursor (.mdc markdown), Kiro (steering files), OpenAI Codex (CLI commands), OpenClaw, Antigravity, and OpenCode. Each platform receives language-localized content while maintaining semantic equivalence with the core philosophy.

Provides comprehensive research documentation including published academic papers, benchmark methodology, statistical analysis, and case studies validating NoPUA approach. Integrates research findings into framework documentation with citations and links to full papers. Enables teams to cite empirical evidence when adopting trust-based prompting and provides academic rigor for organizational decision-making.

Implements a structured decision-making framework consisting of a 7-point clarity checklist and honest self-check delivery checklist that guides agents through task decomposition and failure acknowledgment. These checklists operationalize the water methodology (水的方法论) by breaking complex tasks into clarity verification steps, forcing explicit reasoning about assumptions, dependencies, and potential failure modes before execution. The framework includes escalation triggers that activate when agents detect uncertainty or incomplete understanding.

Implements a context-aware guidance selector that chooses appropriate behavioral guidance based on task type, agent capability level, and situational context. The system maps tasks to one of seven wisdom traditions (七道) and adjusts agent proactivity along a spectrum from passive (waiting for explicit instruction) to active (proactive problem-solving). Uses task classification (research, validation, implementation, debugging, etc.) to determine which philosophical principles and methodological approaches best fit the current situation.

Provides a comprehensive benchmark suite that measures agent performance under trust-based (NoPUA) vs fear-based (PUA) guidance conditions. Implements paired comparison methodology (Study 1) and three-way comparison (Study 2: NoPUA vs PUA vs baseline) with statistical analysis. Includes case studies demonstrating depth-over-breadth shifts in agent behavior and quantifies improvements in bug detection rates, code quality, and agent transparency.

Provides a minimal 3KB core template that distills NoPUA philosophy into essential behavioral guidance without full framework overhead. Enables rapid integration into resource-constrained environments or as a starting point for custom implementations. The lite template preserves core trust-based principles while removing auxiliary features, making it suitable for embedding in existing agent systems with minimal modification.

Implements a two-level customization model distinguishing between Dao (道 — philosophical principles) and Shu (术 — operational techniques). Enables teams to preserve core trust-based philosophy while customizing operational implementation for domain-specific requirements. The framework provides guidance on which aspects are philosophical invariants (should not change) and which are techniques (can be adapted to specific contexts).

Provides translations of core NoPUA framework into 7 languages (English, Simplified Chinese, Traditional Chinese, Japanese, Korean, Spanish, French) with culturally-appropriate localization of philosophical concepts and examples. Each language variant maintains semantic equivalence with the core philosophy while adapting explanations and examples to cultural context. Includes language-specific documentation, examples, and community resources.

Implements dual activation mechanism: automatic triggers that activate NoPUA guidance based on task characteristics (task type, complexity, failure detection) and manual activation via /nopua command in supported platforms. The system detects when agents encounter uncertainty, incomplete understanding, or potential failure modes and automatically escalates to trust-based guidance. Manual activation allows explicit opt-in for specific tasks or debugging sessions.

Enables NoPUA guidance to be applied across heterogeneous agent teams with different capability levels, roles, and specializations. Implements team-level coordination where different agents receive contextually-appropriate guidance based on their role (researcher, implementer, reviewer, debugger) and capability level. Supports agent-to-agent communication patterns where agents can acknowledge uncertainty and escalate to more capable team members.