Skill

reflection

A recursive pattern where an agent evaluates and critiques its own output to iteratively improve quality and catch errors. Use when user asks to "add self-reflection", "agent introspection", "self-critique", or mentions self-evaluation, meta-cognition, or quality self-assessment.

Install

npx claudepluginhub lauraflorentin/skills-marketplace --plugin agentic-skills

Tool Access

This skill uses the workspace's default tool permissions.

Preview

Reflection (also known as Self-Correction or Self-Refinement) gives an agent the ability to look at its work and ask, "Is this correct?" or "Can this be better?". Instead of accepting the first draft, the agent acts as its own critic, identifying flaws and generating an improved version. This significantly boosts performance on reasoning and coding tasks.

SKILL.md

Similar Skills

cache-components

Guides Next.js Cache Components and Partial Prerendering (PPR) with cacheComponents enabled. Implements 'use cache', cacheLife(), cacheTag(), revalidateTag(), static/dynamic optimization, and cache debugging.

cache-components

139.0k

claude-opus-4-5-migration

2 files

Migrates code, prompts, and API calls from Claude Sonnet 4.0/4.5 or Opus 4.1 to Opus 4.5, updating model strings on Anthropic, AWS, GCP, Azure platforms.

claude-opus-4-5-migration

83.2k

bmad-editorial-review-prose

Reviews prose for communication issues impeding comprehension, outputs minimal fixes in a three-column table per Microsoft Writing Style Guide. Useful for 'review prose' or 'improve prose' requests.

bmad-pro-skills

43.8k

Stats

Parent Repo Stars0

Parent Repo Forks0

Last CommitMar 10, 2026

Actions

View Source View Plugin View on GitHub View README

Reflection

When to Use

Quality Control: When high accuracy is required (e.g., generating code, writing legal text).

Compliance: To ensure the output follows specific formatting or policy constraints.

Iterative Improvement: When "good enough" isn't enough, and polish is needed.

Hallucination Check: Asking the model to verify facts against its own knowledge or retrieved context.

Use Cases

Code Repair: Creating code -> Running it -> Reading the error -> Fixing the code.

Writing Polish: Drafting an email -> Critiquing tone and clarity -> Rewriting.

Safety Check: Generating a response -> Checking for policy violations -> Regenerating if unsafe.

Implementation Pattern

def reflection_workflow(task): # Step 1: Draft draft = llm_call(f"Write a draft for: {task}") # Step 2: Critique # The prompt explicitly asks to find problems, not just "rewrite". critique = llm_call( prompt="Critique this draft. List logical errors, tone issues, or missing info.", input=draft ) # Step 3: Refine # The final generation sees both the draft and the specific feedback. final_version = llm_call( prompt="Rewrite the draft to address the following critique...", input={"draft": draft, "critique": critique} ) return final_version

Reflection

When to Use

Quality Control: When high accuracy is required (e.g., generating code, writing legal text).
Compliance: To ensure the output follows specific formatting or policy constraints.
Iterative Improvement: When "good enough" isn't enough, and polish is needed.
Hallucination Check: Asking the model to verify facts against its own knowledge or retrieved context.

Use Cases

Code Repair: Creating code -> Running it -> Reading the error -> Fixing the code.
Writing Polish: Drafting an email -> Critiquing tone and clarity -> Rewriting.
Safety Check: Generating a response -> Checking for policy violations -> Regenerating if unsafe.

Implementation Pattern

def reflection_workflow(task):
    # Step 1: Draft
    draft = llm_call(f"Write a draft for: {task}")
    
    # Step 2: Critique
    # The prompt explicitly asks to find problems, not just "rewrite".
    critique = llm_call(
        prompt="Critique this draft. List logical errors, tone issues, or missing info.",
        input=draft
    )
    
    # Step 3: Refine
    # The final generation sees both the draft and the specific feedback.
    final_version = llm_call(
        prompt="Rewrite the draft to address the following critique...",
        input={"draft": draft, "critique": critique}
    )
    
    return final_version