proactive-task-validator

Official

Audit proactive tasks with checklist-safe checks.

AuthorSimplified-Reasoning
Version1.0.0
Installs0

System Documentation

What problem does it solve?

This Skill helps you validate proactive personal-assistant dataset tasks end-to-end, ensuring that checklists can be satisfied only with the intended hidden intent and dependency intent context—without leaking checklist information into candidate generation.

Core Features & Use Cases

  • Checklist-safe proactive validation: Verifies three-step pass/fail logic so Step 1 and Step 2 satisfy checklist items, while Step 3 (prompt-only) should fail.
  • Dependency-group isolation: Builds validation packets per dependency group to prevent checklist leakage across tasks.
  • Two-stage workflow orchestration: Generates isolated candidate outputs via subagents and performs centralized, itemized checklist judging.
  • Dataset audit reporting: Produces a reviewed Markdown report with structural warnings, per-task verdicts, and repair guidance for weak tasks.

Quick Start

Use the proactive-task-validator skill to validate a dataset rooted at a folder containing episode.yaml and tasks/<task_id>/task.yaml by asking it to run on that dataset root and produce a reviewed report with filled verdicts.

Dependency Matrix

Required Modules

pyyamlpython3jsonpathlibargparse

Components

scripts

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: proactive-task-validator
Download link: https://github.com/Simplified-Reasoning/Pi-Bench/archive/main.zip#proactive-task-validator

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 471,000+ vetted skills library on demand.