skills.rest
qf-studio
qf-studioOFFICIAL
Explore all skills by this author →
WebsiteGitHub

nav-brief

Renders a one-screen intent brief confirming scope before implementing ambiguous task prompts.

232|12|Updated Oct 10, 2025
Search Tags:#drift detection#claude code#ambiguity detection#triz#intent brief#scope confirmation#prompt hook
One-click install
npx skills add https://github.com/qf-studio/navigator --skill nav-brief-qf-studio
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: nav-brief
Source: https://github.com/qf-studio/navigator/tree/main/skills/nav-brief
Command: npx skills add https://github.com/qf-studio/navigator --skill nav-brief-qf-studio

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Ambiguous task prompts like "fix the bug" or "refactor the API" cause AI assistants to implement on guessed scope, burning tokens on rework and polluting context with dead ends. This Skill front-loads scope negotiation by rendering a structured intent brief and waiting for confirmation before any file changes. ## Core Features & Use Cases - Intent Brief Rendering: Produces a one-screen brief covering Goal, Scope, Approach, Limits, Verify, Won't-do, and Contradiction fields, pre-filled from memory and session context with at most 2 open questions. - Hook-Gated Triggering: Fires only when the prompt_brief hook injects a NAV-BRIEF reminder based on an ambiguity score threshold, keeping behavior predictable and tunable via config. - Brief Drift Detection: Mid-implementation, stops and surfaces a warning when work is about to exceed the confirmed brief's scope or limits. - TRIZ Contradiction Handling: Queries the knowledge graph for prior resolutions when a task involves a genuine trade-off, and hands off substantial contradictions to nav-triz. - Use Case: A user types "add rate limiting to the API". The hook scores the prompt as ambiguous, the brief renders with scope pre-filled to public routes and one open question about limit numbers, and implementation starts only after the user confirms. ## Quick Start Ask the assistant to implement an ambiguous task such as "refactor the API" and confirm or edit the intent brief it renders before any code changes begin.

Frequently Asked Questions about nav-brief

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I confirm task scope before an AI writes code?▼

Use an intent brief that renders Goal, Scope, Approach, Limits, Verify, and Won't-do fields before any file edits. The assistant pre-fills defaults from memory, asks at most 2 open questions, and waits for your confirmation before modifying the project.

What triggers the intent brief to appear?▼

The brief renders only when the prompt_brief hook injects a NAV-BRIEF reminder on UserPromptSubmit. The hook fires when the ambiguity scorer rates the prompt at or above the configured threshold, which defaults to 0.5 in the brief_hook config.

How is prompt ambiguity scored?▼

The ambiguity scorer checks for task verbs, vague object references, and undefined dimensions like scope, limits, approach, and verification. Questions, confirmations, and non-task prompts score zero, while file-scoped or acceptance-criteria prompts score below threshold.

Can I skip the brief for urgent fixes?▼

Yes. Signals like "just do it", "quick fix", or "skip the brief" are respected silently and the brief is bypassed. Prompts that answer a pending brief's open questions also pass through without re-rendering.

What happens if implementation exceeds the confirmed scope?▼

The skill raises a BRIEF DRIFT warning naming what exceeds the confirmed brief, then offers two options: extend the brief or skip the change. Work pauses until you answer, preventing silent scope expansion.

What are the limitations of the brief hook?▼

Version 1 is stateless, so it may re-fire on answers containing task verbs, and it lacks a strict_block mode to hard-gate implementation. Measurement counters and prompt rewriting are explicitly deferred to future versions.

Related Skills

Explore More Software Engineering Skills→
Yoodaddy0311Yoodaddy0311
3

cognitive-routing

Routes user requests between fast pattern matching and deep deliberative reasoning using complexity scoring.

Community
Advanced
tilework-techtilework-tech
149

Memorize

Save implementation decisions and patterns to the Nori knowledge base via CLI with name and content parameters.

Official
Intermediate
bketelsenbketelsen
1

memory-search

Search stored memory across frontmatter fields, full-text content, and cross-reference graphs.

Community
Intermediate
Leap0920Leap0920
1

deep-interview

Run Socratic one-question-at-a-time clarification with ambiguity scoring for product ideas.

Community
Advanced