open-medica/open-medical-skills · GitHub

cognitive-behavioral-therapy-tools

54회 설치

어떤 스킬인가요?

Evidence-based cognitive behavioral therapy (CBT) tools and worksheets. Includes thought records, behavioral activation, and cognitive restructuring exercises.

설치 방법

프로젝트 폴더에서 아래 명령어를 실행하고, 설치할 에이전트를 선택하세요.

npx skills add open-medica/open-medical-skills --skill cognitive-behavioral-therapy-tools

설치 명령을 직접 실행해야 적용됩니다. 지원 에이전트와 필요한 권한·라이선스는 제작자의 안내를 확인하세요.

스킬 원문 파일

`instructions.md`

# CBT Therapeutic Techniques Guide

## Overview

The CBT Therapeutic Techniques Guide provides structured, evidence-based cognitive behavioral therapy tools for clinicians and patients. Rooted in Aaron Beck's cognitive model and David Burns' cognitive distortion taxonomy, this skill implements core CBT interventions including thought records, cognitive restructuring, behavioral activation planning, and cognitive distortion identification.

## Clinical Context

Cognitive Behavioral Therapy is one of the most extensively researched psychotherapeutic modalities, with strong evidence for the treatment of major depressive disorder, generalized anxiety disorder, panic disorder, social anxiety, PTSD, OCD, insomnia, and chronic pain (Hofmann et al., 2012). CBT works by identifying and modifying maladaptive thought patterns and behaviors that maintain psychological distress.

This skill serves as a structured adjunct to therapy, providing standardized worksheets and psychoeducational content that clinicians can use during sessions or assign as homework.

## Tools Included

- **Thought Record:** Implements the standard 7-column CBT thought record for identifying automatic thoughts, examining evidence, and generating balanced alternative thoughts.
- **Cognitive Distortion Identifier:** Analyzes automatic thoughts and flags potential cognitive distortions from Burns' taxonomy (14 distortion types), with challenge questions for each.
- **Behavioral Activation Planner:** Generates graded activity schedules for patients with depression, incorporating mastery and pleasure activities with mood monitoring.
- **Distortion Reference Guide:** Complete reference of all cognitive distortions with definitions, clinical examples, and Socratic challenge questions.

## Usage

Clinicians can use these tools during therapy sessions to structure cognitive restructuring exercises, or assign them as between-session homework. The distortion identifier can serve as a teaching tool for patients learning to recognize their own thinking patterns.

## Safety Considerations

This tool does not constitute psychotherapy and does not replace the therapeutic relationship with a licensed mental health professional. Patients experiencing suicidal ideation should be directed to emergency services or crisis resources immediately.

## References

- Beck AT. *Cognitive Therapy and the Emotional Disorders*. Penguin Books, 1976.
- Burns DD. *Feeling Good: The New Mood Therapy*. William Morrow, 1980.
- Hofmann SG, et al. "The Efficacy of CBT: A Review of Meta-Analyses." *Cogn Ther Res*. 2012;36(5):427-440.
- Cuijpers P, et al. "A Meta-Analysis of CBT for Adult Depression." *J Consult Clin Psychol*. 2013;81(3):456-467.

`script.py`

#!/usr/bin/env python3
"""
CBT Therapeutic Techniques Guide
=================================
Evidence-based Cognitive Behavioral Therapy tools implementing core CBT
techniques including thought records, cognitive restructuring, behavioral
activation scheduling, and cognitive distortion identification.

Clinical Purpose:
    Provides structured CBT worksheets and exercises for clinicians and
    patients. Implements Beck's cognitive model with standardized thought
    record formats, common cognitive distortion taxonomy, and behavioral
    activation planning tools.

References:
    - Beck AT. Cognitive Therapy and the Emotional Disorders. 1976.
    - Burns DD. Feeling Good: The New Mood Therapy. 1980.
    - Hofmann SG, et al. "The Efficacy of CBT: A Review of Meta-Analyses."
      Cognitive Therapy and Research. 2012;36(5):427-440.
    - Cuijpers P, et al. "A Meta-Analysis of CBT for Adult Depression."
      J Consult Clin Psychol. 2013;81(3):456-467.

DISCLAIMER: This tool is for clinical decision support and patient education
only. It does not constitute psychotherapy and does not replace the
therapeutic relationship with a licensed mental health professional.
"""

import json
from dataclasses import dataclass, field
from datetime import datetime
from typing import Optional
from enum import Enum


class CognitiveDistortion(Enum):
    """
    Common cognitive distortions (Burns, 1980; Beck, 1976).
    These are systematic errors in thinking that maintain negative beliefs.
    """
    ALL_OR_NOTHING = "all_or_nothing_thinking"
    OVERGENERALIZATION = "overgeneralization"
    MENTAL_FILTER = "mental_filter"
    DISQUALIFYING_POSITIVE = "disqualifying_the_positive"
    JUMPING_TO_CONCLUSIONS = "jumping_to_conclusions"
    MIND_READING = "mind_reading"
    FORTUNE_TELLING = "fortune_telling"
    MAGNIFICATION = "magnification_or_minimization"
    EMOTIONAL_REASONING = "emotional_reasoning"
    SHOULD_STATEMENTS = "should_statements"
    LABELING = "labeling"
    PERSONALIZATION = "personalization"
    CATASTROPHIZING = "catastrophizing"
    BLAME = "blame"


DISTORTION_DESCRIPTIONS = {
    CognitiveDistortion.ALL_OR_NOTHING: {
        "name": "All-or-Nothing Thinking",
        "description": "Seeing things in black-and-white categories. If performance falls short of perfect, you see yourself as a total failure.",
        "example": "I didn't get an A, so I'm a complete failure.",
        "challenge": "Is there a middle ground? Can something be partially successful?",
    },
    CognitiveDistortion.OVERGENERALIZATION: {
        "name": "Overgeneralization",
        "description": "Seeing a single negative event as a never-ending pattern of defeat.",
        "example": "I got rejected once; I'll always be rejected.",
        "challenge": "Is this truly always the case, or is this one instance?",
    },
    CognitiveDistortion.MENTAL_FILTER: {
        "name": "Mental Filter",
        "description": "Picking out a single negative detail and dwelling on it exclusively.",
        "example": "I got one negative comment on my presentation even though 20 were positive.",
        "challenge": "What is the full picture? What positives am I filtering out?",
    },
    CognitiveDistortion.CATASTROPHIZING: {
        "name": "Catastrophizing",
        "description": "Expecting the worst possible outcome or magnifying the importance of negative events.",
        "example": "If I fail this exam, my entire career is over.",
        "challenge": "What is the most likely outcome? What would I tell a friend?",
    },
    CognitiveDistortion.EMOTIONAL_REASONING: {
        "name": "Emotional Reasoning",
        "description": "Assuming that negative emotions reflect the way things really are.",
        "example": "I feel anxious, so something bad must be about to happen.",
        "challenge": "Feelings are not facts. What evidence exists beyond my emotions?",
    },
    CognitiveDistortion.SHOULD_STATEMENTS: {
        "name": "Should Statements",
        "description": "Motivating yourself with 'shoulds' and 'musts,' leading to guilt and frustration.",
        "example": "I should always be productive. I must never make mistakes.",
        "challenge": "Is this a preference or a rigid demand? What happens if I replace 'should' with 'prefer'?",
    },
    CognitiveDistortion.PERSONALIZATION: {
        "name": "Personalization",
        "description": "Seeing yourself as the cause of a negative event you were not primarily responsible for.",
        "example": "My patient didn't improve; it must be because I'm a bad clinician.",
        "challenge": "What other factors contributed? Am I taking responsibility for things outside my control?",
    },
    CognitiveDistortion.MIND_READING: {
        "name": "Mind Reading",
        "description": "Assuming you know what others are thinking without evidence.",
        "example": "My supervisor thinks I'm incompetent.",
        "challenge": "Do I have concrete evidence for this? Have I asked directly?",
    },
    CognitiveDistortion.FORTUNE_TELLING: {
        "name": "Fortune Telling",
        "description": "Predicting negative outcomes without evidence.",
        "example": "I'm going to fail the interview tomorrow.",
        "challenge": "Can I really predict the future? What evidence supports a different outcome?",
    },
    CognitiveDistortion.LABELING: {
        "name": "Labeling",
        "description": "Attaching a global label to yourself or others based on a single event.",
        "example": "I made a mistake, so I'm an idiot.",
        "challenge": "Am I defining myself by one event? Would I label someone else this way?",
    },
    CognitiveDistortion.MAGNIFICATION: {
        "name": "Magnification/Minimization",
        "description": "Exaggerating the importance of negative events or minimizing positive ones.",
        "example": "My small error is a disaster, but my promotion was just luck.",
        "challenge": "Am I keeping things in proportion? How significant will this be in a year?",
    },
    CognitiveDistortion.DISQUALIFYING_POSITIVE: {
        "name": "Disqualifying the Positive",
        "description": "Rejecting positive experiences by insisting they don't count.",
        "example": "They only said I did well to be nice.",
        "challenge": "Why can't positive feedback be genuine? What evidence contradicts my dismissal?",
    },
    CognitiveDistortion.JUMPING_TO_CONCLUSIONS: {
        "name": "Jumping to Conclusions",
        "description": "Making negative interpretations without definite facts to support the conclusion.",
        "example": "She didn't reply to my text; she must be angry with me.",
        "challenge": "What are other possible explanations?",
    },
    CognitiveDistortion.BLAME: {
        "name": "Blame",
        "description": "Holding other people entirely responsible for your distress, or blaming yourself for everything.",
        "example": "It's all their fault that I feel this way.",
        "challenge": "What is within my control? What responsibility belongs elsewhere?",
    },
}


@dataclass
class ThoughtRecord:
    """
    Standard 7-column CBT thought record (Beck, 1995).
    Used to identify, evaluate, and restructure automatic thoughts.
    """
    record_id: str = ""
    date: str = ""
    situation: str = ""
    emotions: list = field(default_factory=list)  # [{"emotion": str, "intensity": 0-100}]
    automatic_thoughts: list = field(default_factory=list)
    cognitive_distortions: list = field(default_factory=list)
    evidence_for: list = field(default_factory=list)
    evidence_against: list = field(default_factory=list)
    balanced_thought: str = ""
    new_emotion_rating: list = field(default_factory=list)  # [{"emotion": str, "intensity": 0-100}]

    def __post_init__(self):
        if not self.date:
            self.date = datetime.now().strftime("%Y-%m-%d %H:%M")
        if not self.record_id:
            self.record_id = f"TR-{datetime.now().strftime('%Y%m%d%H%M%S')}"


@dataclass
class BehavioralActivationEntry:
    """
    Behavioral activation schedule entry.
    Tracks activity, predicted vs actual mood, and mastery/pleasure ratings.
    """
    date: str
    time_block: str  # morning, afternoon, evening
    planned_activity: str
    activity_type: str  # mastery, pleasure, necessary
    predicted_mood: int = 5  # 0-10
    actual_mood: Optional[int] = None
    mastery_rating: Optional[int] = None  # 0-10
    pleasure_rating: Optional[int] = None  # 0-10
    completed: bool = False


def identify_distortions(thought: str) -> list:
    """
    Analyze an automatic thought and identify potential cognitive distortions.

    This uses keyword pattern matching as a screening aid. Clinical judgment
    should always be applied to the final interpretation.
    """
    thought_lower = thought.lower()
    identified = []

    # All-or-nothing indicators
    if any(w in thought_lower for w in ["always", "never", "nothing", "everything",
                                         "completely", "totally", "perfect"]):
        identified.append(CognitiveDistortion.ALL_OR_NOTHING)

    # Overgeneralization
    if any(w in thought_lower for w in ["always", "never", "everyone", "no one",
                                         "every time", "nothing ever"]):
        identified.append(CognitiveDistortion.OVERGENERALIZATION)

    # Catastrophizing
    if any(w in thought_lower for w in ["worst", "terrible", "disaster", "horrible",
                                         "catastrophe", "end of", "ruined"]):
        identified.append(CognitiveDistortion.CATASTROPHIZING)

    # Should statements
    if any(w in thought_lower for w in ["should", "must", "ought to", "have to",
                                         "supposed to"]):
        identified.append(CognitiveDistortion.SHOULD_STATEMENTS)

    # Mind reading
    if any(w in thought_lower for w in ["they think", "he thinks", "she thinks",
                                         "everyone thinks", "they must think",
                                         "probably thinks"]):
        identified.append(CognitiveDistortion.MIND_READING)

    # Emotional reasoning
    if any(w in thought_lower for w in ["i feel like", "i feel that",
                                         "it feels like", "feels wrong"]):
        identified.append(CognitiveDistortion.EMOTIONAL_REASONING)

    # Labeling
    if any(w in thought_lower for w in ["i'm a ", "i am a ", "i'm such a",
                                         "idiot", "failure", "loser", "worthless"]):
        identified.append(CognitiveDistortion.LABELING)

    # Personalization
    if any(w in thought_lower for w in ["my fault", "because of me",
                                         "i caused", "blame myself"]):
        identified.append(CognitiveDistortion.PERSONALIZATION)

    # Fortune telling
    if any(w in thought_lower for w in ["going to fail", "will never", "won't work",
                                         "bound to", "doomed"]):
        identified.append(CognitiveDistortion.FORTUNE_TELLING)

    # Remove duplicates while preserving order
    seen = set()
    unique = []
    for d in identified:
        if d not in seen:
            seen.add(d)
            unique.append(d)

    return unique


def create_thought_record(situation: str, emotions: list,
                          automatic_thoughts: list) -> dict:
    """
    Create a structured CBT thought record and guide the restructuring process.

    Args:
        situation: Description of the triggering situation
        emotions: List of dicts [{"emotion": str, "intensity": int}]
        automatic_thoughts: List of automatic thoughts (hot thoughts)

    Returns:
        Structured thought record with distortion analysis and restructuring prompts
    """
    record = ThoughtRecord(
        situation=situation,
        emotions=emotions,
        automatic_thoughts=automatic_thoughts,
    )

    # Identify cognitive distortions across all automatic thoughts
    all_distortions = []
    thought_analysis = []
    for thought in automatic_thoughts:
        distortions = identify_distortions(thought)
        distortion_details = []
        for d in distortions:
            info = DISTORTION_DESCRIPTIONS.get(d, {})
            distortion_details.append({
                "distortion": info.get("name", d.value),
                "description": info.get("description", ""),
                "challenge_question": info.get("challenge", ""),
            })
            all_distortions.append(d)

        thought_analysis.append({
            "thought": thought,
            "identified_distortions": distortion_details,
        })

    record.cognitive_distortions = list(set(all_distortions))

    # Generate restructuring prompts
    restructuring_prompts = [
        "What is the evidence FOR this thought?",
        "What is the evidence AGAINST this thought?",
        "Is there an alternative explanation?",
        "What is the worst that could happen? Could I survive it?",
        "What is the best that could happen?",
        "What is the MOST REALISTIC outcome?",
        "What would I tell a close friend who had this thought?",
        "What is the effect of believing this thought? Of changing my thinking?",
    ]

    return {
        "thought_record": {
            "record_id": record.record_id,
            "date": record.date,
            "situation": record.situation,
            "emotions": record.emotions,
            "automatic_thoughts": thought_analysis,
            "restructuring_prompts": restructuring_prompts,
            "next_steps": [
                "1. Examine the evidence for and against each automatic thought",
                "2. Generate a balanced/alternative thought",
                "3. Re-rate your emotion intensity (0-100) after restructuring",
                "4. Note any changes in how you feel",
            ],
        },
        "disclaimer": "This is a clinical support tool, not a substitute for psychotherapy.",
    }


def generate_behavioral_activation_plan(current_mood: int,
                                        interests: list,
                                        barriers: list = None) -> dict:
    """
    Generate a behavioral activation schedule.

    Behavioral activation is an evidence-based treatment for depression that
    focuses on increasing engagement in valued activities to improve mood.

    Args:
        current_mood: Self-rated mood (0-10, where 10 is best)
        interests: List of activities the patient enjoys or used to enjoy
        barriers: List of perceived barriers to activity

    Returns:
        Structured weekly activity schedule with graded activities
    """
    barriers = barriers or []

    # Activity difficulty grading based on current mood
    if current_mood <= 3:
        difficulty = "gentle"
        activity_duration = "15-30 minutes"
        daily_target = 2
        guidance = ("Start with very small, manageable activities. The goal is "
                    "simply to begin moving, not to achieve a specific outcome. "
                    "Even getting dressed or taking a 5-minute walk counts.")
    elif current_mood <= 6:
        difficulty = "moderate"
        activity_duration = "30-60 minutes"
        daily_target = 3
        guidance = ("Build on small successes with moderately engaging activities. "
                    "Mix necessary tasks with pleasurable ones. Notice the "
                    "connection between activity and mood.")
    else:
        difficulty = "full"
        activity_duration = "60+ minutes"
        daily_target = 4
        guidance = ("Maintain your activity level and add new challenges. "
                    "Focus on activities that align with your values and "
                    "provide both mastery and pleasure.")

    # Categorize activities
    mastery_activities = [
        "Complete one household task",
        "Prepare a healthy meal",
        "Organize one small area",
        "Learn something new for 15 minutes",
        "Practice a skill or hobby",
    ]

    pleasure_activities = [
        "Listen to music you enjoy",
        "Spend time in nature (even 10 minutes)",
        "Connect with a friend or family member",
        "Read something enjoyable",
        "Practice mindful relaxation",
    ]

    # Add personalized activities from interests
    for interest in interests[:5]:
        pleasure_activities.append(f"Engage in {interest} (even briefly)")

    schedule = {
        "difficulty_level": difficulty,
        "activity_duration": activity_duration,
        "daily_target": daily_target,
        "guidance": guidance,
        "mastery_activities": mastery_activities,
        "pleasure_activities": pleasure_activities,
        "monitoring_instructions": [
            "Before each activity: rate your predicted mood (0-10)",
            "After each activity: rate your actual mood (0-10)",
            "Note: actual mood often exceeds predicted mood",
            "Rate mastery (sense of accomplishment, 0-10) and pleasure (0-10)",
            "Review patterns at end of week with your therapist",
        ],
    }

    if barriers:
        schedule["barrier_strategies"] = {
            barrier: "Break the activity into the smallest possible step. "
                     "Commit to just 2 minutes. You can always stop after 2 minutes."
            for barrier in barriers
        }

    return {
        "behavioral_activation_plan": schedule,
        "disclaimer": "Behavioral activation should be guided by a licensed therapist.",
    }


def run_cbt_tool(tool: str, **kwargs) -> str:
    """
    Main entry point for CBT tools.

    Tools:
        thought_record: Create a thought record
        identify_distortions: Analyze a thought for cognitive distortions
        behavioral_activation: Generate an activation plan
        distortion_reference: Get reference for all cognitive distortions
    """
    if tool == "thought_record":
        result = create_thought_record(
            kwargs.get("situation", ""),
            kwargs.get("emotions", []),
            kwargs.get("automatic_thoughts", []),
        )
    elif tool == "identify_distortions":
        thought = kwargs.get("thought", "")
        distortions = identify_distortions(thought)
        result = {
            "thought": thought,
            "distortions": [
                DISTORTION_DESCRIPTIONS[d] for d in distortions
            ],
        }
    elif tool == "behavioral_activation":
        result = generate_behavioral_activation_plan(
            kwargs.get("current_mood", 5),
            kwargs.get("interests", []),
            kwargs.get("barriers", []),
        )
    elif tool == "distortion_reference":
        result = {
            "cognitive_distortions": {
                d.value: DISTORTION_DESCRIPTIONS[d]
                for d in CognitiveDistortion
                if d in DISTORTION_DESCRIPTIONS
            },
        }
    else:
        result = {
            "error": f"Unknown tool: {tool}",
            "available_tools": ["thought_record", "identify_distortions",
                                "behavioral_activation", "distortion_reference"],
        }

    return json.dumps(result, indent=2)


if __name__ == "__main__":
    print("=== Thought Record Example ===")
    print(run_cbt_tool(
        "thought_record",
        situation="Presented at morning conference and stumbled over a question",
        emotions=[{"emotion": "anxiety", "intensity": 75},
                  {"emotion": "shame", "intensity": 60}],
        automatic_thoughts=[
            "Everyone thinks I'm incompetent",
            "I should always know the answer",
            "My career is going to fail because of this",
        ],
    ))

`SKILL.md`

---
name: cognitive-behavioral-therapy-tools
description: >
Evidence-based cognitive behavioral therapy (CBT) tools and worksheets.
Includes thought records, behavioral activation, and cognitive
restructuring exercises.
---