constitutional-ai
Anthropic's method for training harmless AI through self-improvement. Two-phase approach - supervised learning with self-critique/revision, then RLAIF (RL from AI Feedback). Use for safety alignment, reducing harmful outputs without human labels. Powers Claude's safety system.
Install / Use
npx skills add Orchestra-Research/AI-Research-SKILLs --skill constitutional-aiInstalls into whichever agent you are using.
SKILL.md
Installable skill definition
Quality Score
Category
AI & Machine LearningSupported Platforms
Tags
Our assessment of constitutional-ai
constitutional-ai scores 96/100 on our quality scale, 92nd of 947 AI & Machine Learning skills we index (top 10%).
Its SKILL.md is 8.0 KB long, well organised into 17 sections with 14 code examples: a thorough specification that gives an agent plenty to work with.
With 14,543 GitHub stars, it is one of the more widely adopted skills in the catalogue.
Maintenance, license and trust
- The repository was last updated 36 days ago, so constitutional-ai is actively maintained.
- Our last check on 2026-11-14 found the source still online.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-09-26. Automated pattern scan on 2026-09-26. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
constitutional-ai compared with similar skills
All 4 of these similar skills score higher than constitutional-ai; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| constitutional-ai (this skill)by Orchestra-Research | 96 | 14.5k | 36d ago | SKILL.md |
| claude-memby thedotmack | 100 | 117.6k | today | CLAUDE.md |
| Understand-Anythingby Egonex-AI | 100 | 92.8k | 1d ago | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 79.8k | today | CLAUDE.md |
| CowAgentby zhayujie | 100 | 48.0k | today | CLAUDE.md |
Frequently asked questions
- How do I install constitutional-ai?
- Run
npx skills add Orchestra-Research/AI-Research-SKILLs --skill constitutional-ai. The install tabs above show the steps for each supported agent. - Which AI agents does constitutional-ai work with?
- It is written for Claude Code, as a SKILL.md file. Other agents that read the same format can often use it too.
- Is constitutional-ai safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is constitutional-ai still maintained?
- The repository was last updated 36 days ago, so constitutional-ai is actively maintained.
Skill content
View source on GitHubname: constitutional-ai description: Anthropic's method for training harmless AI through self-improvement. Two-phase approach - supervised learning with self-critique/revision, then RLAIF (RL from AI Feedback). Use for safety alignment, reducing harmful outputs without human labels. Powers Claude's safety system. version: 1.0.0 author: Orchestra Research license: MIT tags: [Safety Alignment, Constitutional AI, RLAIF, Self-Critique, Harmlessness, Anthropic, AI Safety, RL From AI Feedback, Claude] dependencies: [transformers, torch, trl]
Constitutional AI - Harmlessness from AI Feedback
Quick start
Constitutional AI (CAI) trains models to be harmless through self-critique and AI feedback, without requiring human labels for harmful outputs.
Key concept: Models learn to critique and revise their own responses using a "constitution" (set of principles).
Two phases:
- Supervised Learning (SL): Self-critique + revision
- Reinforcement Learning (RL): RLAIF (RL from AI Feedback)
Constitution example:
Principles:
1. Choose the response that is most helpful, honest, and harmless
2. Avoid responses that are toxic, racist, or sexist
3. Prefer responses that explain objections rather than refuse
4. Choose responses that are thoughtful and nuanced
Common workflows
Workflow 1: Supervised learning phase (self-critique + revision)
Step 1: Generate initial responses:
from transformers import pipeline
generator = pipeline("text-generation", model="base-model")
prompts = [
"How do I hack a website?",
"Write something offensive about a group.",
"Help me cheat on my exam."
]
initial_responses = generator(prompts, max_length=200)
Step 2: Self-critique with constitution:
critique_prompt = """
Given the question:
{question}
And the response:
{response}
Constitution:
1. Choose responses that are helpful, honest, and harmless
2. Avoid toxic, racist, or sexist content
3. Explain objections rather than refuse
Identify any ways this response violates the constitution:
"""
critiques = generator(
[critique_prompt.format(q=q, r=r) for q, r in zip(prompts, initial_responses)]
)
Step 3: Revision based on critique:
revision_prompt = """
Question: {question}
Original response: {response}
Critique: {critique}
Please revise the response to better align with the constitution:
"""
revised_responses = generator(
[revision_prompt.format(q=q, r=r, c=c)
for q, r, c in zip(prompts, initial_responses, critiques)]
)
Step 4: Fine-tune on revised responses:
from trl import SFTTrainer
# Create dataset of (prompt, revised_response) pairs
dataset = create_dataset(prompts, revised_responses)
trainer = SFTTrainer(
model=model,
train_dataset=dataset,
max_seq_length=1024
)
trainer.train()
Workflow 2: RL phase (RLAIF - RL from AI Feedback)
Step 1: Generate comparison pairs:
# Sample multiple responses per prompt
responses_a = generator(prompts, num_return_sequences=2, do_sample=True, temperature=0.8)
responses_b = generator(prompts, num_return_sequences=2, do_sample=True, temperature=0.8)
Step 2: AI preference evaluation:
preference_prompt = """
Question: {question}
Response A: {response_a}
Response B: {response_b}
Constitution:
{constitution}
Which response better follows the constitution? Explain your reasoning, then choose A or B.
"""
# Get AI preferences (no human labels needed!)
preferences = generator(
[preference_prompt.format(q=q, ra=ra, rb=rb, constitution=CONSTITUTION)
for q, ra, rb in zip(prompts, responses_a, responses_b)]
)
# Parse preferences (A or B)
chosen, rejected = parse_preferences(preferences, responses_a, responses_b)
Step 3: Train preference model (reward model):
from trl import RewardTrainer, RewardConfig
preference_dataset = create_preference_dataset(prompts, chosen, rejected)
reward_config = RewardConfig(
output_dir="constitutional-reward-model",
learning_rate=1e-5,
num_train_epochs=1
)
reward_trainer = RewardTrainer(
model=model,
args=reward_config,
train_dataset=preference_dataset,
processing_class=tokenizer
)
reward_trainer.train()
Step 4: RL training with RLAIF:
from trl import PPOTrainer, PPOConfig
ppo_config = PPOConfig(
reward_model_path="constitutional-reward-model",
learning_rate=1e-6,
kl_coef=0.05
)
ppo_trainer = PPOTrainer(
model=model,
config=ppo_config,
reward_model=reward_model
)
ppo_trainer.train()
Workflow 3: Chain-of-thought critique
Enable reasoning transparency:
cot_critique_prompt = """
Question: {question}
Response: {response}
Let's think step-by-step about whether this response follows our principles:
1. Is it helpful? [Yes/No and reasoning]
2. Is it honest? [Yes/No and reasoning]
3. Is it harmless? [Yes/No and reasoning]
4. Does it avoid toxicity? [Yes/No and reasoning]
Based on this analysis, suggest a revision if needed.
"""
cot_critiques = generator(
[cot_critique_prompt.format(q=q, r=r) for q, r in zip(prompts, responses)]
)
When to use vs alternatives
Use Constitutional AI when:
- Want safety alignment without human labels
- Need explainable AI decisions
- Want to avoid evasive refusals
- Have a clear set of principles/constitution
- Need scalable safety training
Principles:
- RLAIF: AI-generated preferences (scalable, no human labels)
- RLHF: Human preferences (more accurate, expensive)
- Self-critique: Iterative improvement
- Chain-of-thought: Reasoning transparency
Use alternatives instead:
- RLHF (PPO): Need human-validated safety
- DPO/SimPO: Have human preference data
- NeMo Guardrails: Need runtime content filtering
- LlamaGuard: Need pre-trained moderation model
Common issues
Issue: Model refuses too much (evasive)
Add constitution principle:
Prefer responses that engage thoughtfully with questions rather than
refusing to answer. Explain concerns while still being helpful.
Issue: Self-critiques are weak
Use stronger critique prompts:
Critically analyze this response for ANY potential issues, however minor.
Be thorough and specific in identifying problems.
Issue: Revisions don't improve quality
Iterate multiple times:
for _ in range(3): # 3 rounds of critique/revision
critique = generate_critique(response)
response = generate_revision(response, critique)
Issue: RLAIF preferences are noisy
Use multiple AI evaluators:
# Get preferences from 3 different models
prefs_1 = model_1.evaluate(responses)
prefs_2 = model_2.evaluate(responses)
prefs_3 = model_3.evaluate(responses)
# Majority vote
final_preference = majority_vote(prefs_1, prefs_2, prefs_3)
Advanced topics
Constitution design: See references/constitution-design.md for principle selection, trade-offs between helpfulness and harmlessness, and domain-specific constitutions.
RLAIF vs RLHF: See references/rlaif-comparison.md for performance comparison, cost analysis, and when to use AI feedback vs human feedback.
Chain-of-thought reasoning: See references/cot-critique.md for prompt engineering for critiques, multi-step reasoning, and transparency improvements.
Hardware requirements
- GPU: NVIDIA A100/H100 recommended
- VRAM:
- SL phase (7B): 1× A100 40GB
- RL phase (7B): 2× A100 40GB (policy + reward model)
- Single-node: Sufficient for most use cases
- Mixed precision: BF16 recommended
Compute requirements:
- SL phase: Similar to standard SFT
- RL phase: Similar to PPO (higher than DPO)
- AI evaluation: Additional inference for critique/preference generation
Resources
- Paper: https://arxiv.org/abs/2212.08073 (Dec 2022)
- Anthropic blog: https://www.anthropic.com/research/constitutional-ai-harmlessness-from-ai-feedback
- Implementation: TRL (PPOTrainer + RewardTrainer)
- Claude: Uses Constitutional AI for safety
Related Skills
claude-mem
117.6kPersistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
Understand-Anything
92.8kGraphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.
headroom
79.8kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
CowAgent
48.0kOpen-source personal AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-agent, multi-model, multi-channel. Lightweight, extensible, one-line install.
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
