thothctl

AI-Powered Development Lifecycle (AI-DLC)

What You Can Do

ThothCTL’s AI-DLC lets you interact with your infrastructure lifecycle using AI assistants. Instead of memorizing CLI flags, you describe what you need — the AI executes the right commands and explains the results.

Use Case	What You’ll Achieve	Time
Generate infrastructure from intent	Working Terraform/Terragrunt from a description	2 min
Security review with AI agents	Findings + auto-generated fixes + PR decision	3 min
Full DevSecOps pipeline via AI	Cost + security + blast radius in one conversation	5 min
Analyze planned changes	Cost estimate + risk assessment from plan files	3 min
Detect and fix drift	Drifted resources + remediation guidance	2 min
Generate documentation	README, dependency graphs, architecture diagrams	1 min

The AI-DLC Workflow

%%{init: {'theme':'base', 'themeVariables': { 'primaryColor':'#3f51b5','primaryTextColor':'#fff','primaryBorderColor':'#303f9f','lineColor':'#536dfe','secondaryColor':'#536dfe','tertiaryColor':'#fff'}}}%%
graph TB
    A["🤖 AI Assistant<br/>Kiro · Claude · Copilot"] --> B{"Workflow Choice"}
    
    B -->|"Option 1:<br/>AI Orchestrates"| C["📡 MCP Server<br/>AI calls ThothCTL tools"]
    B -->|"Option 2:<br/>Manual + AI Analysis"| D["💻 Direct CLI<br/>Developer runs commands"]
    
    C --> E["⚙️ ThothCTL Engine<br/>generate, scan, check,<br/>ai-review, inventory, workflow"]
    D --> E
    
    E --> F["📊 Results & Reports<br/>JSON, HTML, Findings, SBOM"]
    
    F --> G["🧠 AI Analysis<br/>Insights, Summaries,<br/>Fixes, Decisions"]
    
    G --> H["🚀 Action<br/>Deploy, Fix, Document,<br/>Create PR"]
    
    H --> A
    
    classDef aiStyle fill:#3f51b5,stroke:#5c6bc0,stroke-width:3px,color:#fff
    classDef choiceStyle fill:#f57f17,stroke:#fbc02d,stroke-width:3px,color:#fff
    classDef mcpStyle fill:#0277bd,stroke:#039be5,stroke-width:3px,color:#fff
    classDef cliStyle fill:#2e7d32,stroke:#43a047,stroke-width:3px,color:#fff
    classDef engineStyle fill:#ef6c00,stroke:#fb8c00,stroke-width:3px,color:#fff
    classDef resultsStyle fill:#c2185b,stroke:#e91e63,stroke-width:3px,color:#fff
    classDef analysisStyle fill:#7b1fa2,stroke:#9c27b0,stroke-width:3px,color:#fff
    classDef actionStyle fill:#00695c,stroke:#00897b,stroke-width:3px,color:#fff
    
    class A aiStyle
    class B choiceStyle
    class C mcpStyle
    class D cliStyle
    class E engineStyle
    class F resultsStyle
    class G analysisStyle
    class H actionStyle

Two ways to use it:

Option 1 — AI Orchestrates: Run kiro-cli chat --agent thoth, describe what you need in natural language, the agent executes ThothCTL commands via MCP and explains results.
Option 2 — Manual + AI Analysis: Run thothctl commands yourself, then start kiro-cli chat --agent thoth to analyze results, prioritize findings, and get fix suggestions.

Prerequisites

1. Install ThothCTL

# Install with pip
pip install thothctl

# Or install with pipx (recommended — isolated environment, no conflicts)
pipx install thothctl

# Already installed? Upgrade to latest version
thothctl upgrade

# Or with pip/pipx
pip install --upgrade thothctl
pipx upgrade thothctl

2. Configure MCP (for AI assistant integration)

Add ThothCTL to your AI assistant’s MCP configuration:

=== “Kiro CLI (~/.kiro/settings/mcp.json)”

```json
{
  "mcpServers": {
    "thothctl": {
      "command": "thothctl",
      "args": ["mcp", "server", "--stdio"]
    }
  }
}
```

=== “Claude Code (~/.claude/settings/mcp.json)”

```json
{
  "mcpServers": {
    "thothctl": {
      "command": "thothctl",
      "args": ["mcp", "server", "--stdio"]
    }
  }
}
```

=== “Per-project (.kiro/settings/mcp.json)”

```json
{
  "mcpServers": {
    "thothctl": {
      "command": "thothctl",
      "args": ["mcp", "server", "--stdio"]
    }
  }
}
```

3. (Optional) Install a local LLM for AI Review

# For local AI review without cloud costs
# Install Ollama: https://ollama.ai
ollama pull llama3.1:8b

Verify Setup

thothctl --version         # Should show v0.27+
thothctl mcp status        # Should show server available
ollama list                # (optional) Shows local models

Use Case 1: Generate Infrastructure from Natural Language

Goal: Describe what you need → get working, compliant IaC code.

Step 1: Initialize your project and start the AI agent

# Initialize a project first (sets up scaffold + org rules)
thothctl init project -p my-infrastructure --project-type terraform-terragrunt --space production

# Start Kiro CLI with the ThothCTL agent
cd my-infrastructure
kiro-cli chat --agent thoth

Step 2: Describe your intent

=== “Via Kiro CLI (AI Agent)”

Once inside the `kiro-cli chat --agent thoth` session:

```
You: "Generate a VPC with 3 private subnets, NAT gateway, and flow logs 
      for production in us-east-1"
```

The agent uses ThothCTL MCP tools to generate, validate, and self-correct automatically.

=== “Via CLI (no AI assistant)”

```bash
thothctl generate iac \
  --intent "VPC with 3 private subnets, NAT gateway, and flow logs" \
  --project-type terraform-terragrunt \
  --provider ollama \
  --space production \
  --apply
```

Step 2: ThothCTL generates, validates, and self-corrects

🔄 Generating infrastructure code...
   ├── Loading org rules from .thothcf.toml
   ├── Using scaffold: terraform-terragrunt
   ├── AI generating HCL code...
   ├── Validating with Checkov...
   │   └── ❌ 1 violation: VPC flow logs not enabled
   ├── Self-correcting (iteration 2/5)...
   ├── Re-validating...
   │   └── ✅ 0 violations
   └── Running terraform plan...
       └── ✅ Plan succeeded (7 resources to create)

✅ Generated files:
   stacks/foundation/network/vpc/
   ├── main.tf          (VPC + subnets + NAT + flow logs)
   ├── variables.tf     (region, cidr, environment)
   ├── outputs.tf       (vpc_id, subnet_ids, nat_gateway_ip)
   └── terragrunt.hcl   (backend + provider config)

📊 Estimated cost: $142/month
📐 Architecture diagram: docs/vpc-architecture.mmd

Step 3: Review and deploy

cd stacks/foundation/network/vpc
terragrunt plan    # Verify the generated code
terragrunt apply   # Deploy when satisfied

Supported Project Types

Type	Flag	What Gets Generated
Terraform	`--project-type terraform`	`main.tf`, `variables.tf`, `outputs.tf`
Terragrunt	`--project-type terraform-terragrunt`	Multi-stack with `terragrunt.hcl` per stack
CloudFormation	`--project-type cloudformation`	`template.yaml` with Parameters/Resources/Outputs
CDK v2	`--project-type cdkv2`	TypeScript/Python CDK constructs

Try It

# Requires an AI provider. Choose one:

# Option A: Local Ollama (free, private — install from https://ollama.ai)
ollama pull llama3.1:8b
thothctl generate iac \
  --intent "S3 bucket with encryption and versioning" \
  --project-type terraform \
  --provider ollama

# Option B: AWS Bedrock (requires AWS credentials with Bedrock access)
thothctl generate iac \
  --intent "S3 bucket with encryption and versioning" \
  --project-type terraform \
  --provider bedrock

# Option C: OpenAI (requires OPENAI_API_KEY env var)
thothctl generate iac \
  --intent "S3 bucket with encryption and versioning" \
  --project-type terraform \
  --provider openai

# Add --apply to write files to disk (dry-run by default)

!!! note “Provider Required” generate iac requires an AI provider to produce code. Without --provider, ThothCTL will use the provider configured in .thothcf.toml or prompt you to set one up.

Use Case 2: AI-Powered Security Review

Goal: Get AI agents to analyze your IaC, find security issues, suggest fixes, and decide if a PR is safe.

Step 1: Run AI review on your code

=== “Via Kiro CLI (AI Agent)”

```bash
# Start Kiro CLI with the ThothCTL agent
kiro-cli chat --agent thoth
```

Then in the chat session:
```
You: "Review my Terraform code in ./stacks for security issues"
```

=== “Via CLI (direct)”

```bash
# With local Ollama (free, private)
thothctl ai-review analyze -d ./stacks -p ollama

# With AWS Bedrock (Claude)
thothctl ai-review analyze -d ./stacks -p bedrock --model us.anthropic.claude-sonnet-4-20250514-v1:0

# With OpenAI
thothctl ai-review analyze -d ./stacks -p openai
```

Step 2: Four AI agents analyze in parallel

🔒 Security Agent:
   - IAM role has wildcard s3:* permissions (iam.tf:15) → HIGH
   - S3 bucket allows public access (storage.tf:8) → HIGH
   - RDS not encrypted at rest (database.tf:22) → MEDIUM

🏗️ Architecture Agent:
   - EKS on single AZ (cluster.tf:5) → Consider multi-AZ for HA
   - No autoscaling on node group → Add cluster-autoscaler

🔧 Fix Agent:
   - Generated fix for iam.tf: Replace "s3:*" with ["s3:GetObject", "s3:PutObject"]
   - Generated fix for storage.tf: Add aws_s3_bucket_public_access_block
   - Generated fix for database.tf: Add storage_encrypted = true

⚖️ Decision Agent:
   Verdict: REQUEST_CHANGES (confidence: 0.91)
   Reason: 2 HIGH security findings must be resolved before merge

Step 3: Make a PR decision (optional)

# Dry-run — shows what decision would be posted
thothctl ai-review decide -d ./stacks --pr-number 42 --dry-run

# Post decision to PR (GitHub/GitLab/Azure DevOps)
thothctl ai-review decide -d ./stacks --pr-number 42

Available Providers

Provider	Flag	Cost	Best For
Ollama	`-p ollama`	Free	Local dev, privacy, testing
AWS Bedrock	`-p bedrock`	Pay-per-token	Enterprise, Claude models
OpenAI	`-p openai`	Pay-per-token	GPT-4o, broad availability
Azure OpenAI	`-p azure`	Pay-per-token	Azure compliance requirements

Try It

# Quickest test — uses local Ollama
ollama pull llama3.1:8b
thothctl ai-review analyze -d . -p ollama

Use Case 3: Run DevSecOps Pipeline via AI

Goal: Execute the full DevSecOps lifecycle (cost → structure → inventory → security → blast radius → drift) in one conversation or one command.

Option A: Via Kiro CLI (conversational)

# Start Kiro CLI with the ThothCTL agent
kiro-cli chat --agent thoth

Then in the chat session:

You: "Run the full DevSecOps pipeline on my infrastructure"

AI: [Executes thothctl workflow devsecops --phase all]

📋 Results:
┌───────────┬──────────┬───────────────────────────────┐
│ Phase     │ Status   │ Summary                       │
├───────────┼──────────┼───────────────────────────────┤
│ Plan      │ ✅ Pass  │ Cost: $2,847/mo               │
│ Develop   │ ✅ Pass  │ Structure valid               │
│ Build     │ ✅ Pass  │ 12 modules, 3 outdated        │
│ Test      │ ✅ Pass  │ 0 plan errors                 │
│ Secure    │ ⚠️ Warn  │ 3 HIGH, 5 MEDIUM findings     │
│ Deploy    │ ✅ Pass  │ Blast radius: Medium          │
│ Monitor   │ ✅ Pass  │ 0 drifted resources           │
└───────────┴──────────┴───────────────────────────────┘

⚠️ 3 HIGH severity findings in Secure phase.
Would you like me to show details and fixes?

Option B: Via CLI (for CI/CD)

# Full pipeline (soft enforcement — report only)
thothctl workflow devsecops --phase all

# Pre-deploy gate (hard enforcement — fails on violations)
thothctl workflow devsecops --phase pre-deploy --enforcement hard

# Single phase
thothctl workflow devsecops --phase secure

Option C: Custom YAML workflow

# Run a custom workflow DAG
thothctl workflow run --file .thothcf_workflow.yaml

# Dry-run to see execution plan
thothctl workflow run --file .thothcf_workflow.yaml --dry-run

Available Phases

Phase	What It Does	CLI Flag
Plan	Cost analysis + blast radius	`--phase plan`
Develop	Environment check + project structure	`--phase develop`
Build	Inventory + version checks (SBOM)	`--phase build`
Test	Terraform plan validation	`--phase test`
Secure	Multi-tool scanning (Checkov, Trivy, OPA)	`--phase secure`
Deploy	Security gate (hard enforcement)	`--phase deploy`
Monitor	Drift detection	`--phase monitor`
Pre-deploy	Test + Secure combined (CI/CD gate)	`--phase pre-deploy`
All	Full pipeline	`--phase all`

Try It

# Quickest test — just the security phase
thothctl workflow devsecops --phase secure

# Full pipeline
thothctl workflow devsecops --phase all

📖 For detailed phase descriptions, plan generation per framework, and CI/CD integration, see the DevSecOps SDLC Guide.

Use Case 4: Analyze Planned Changes (Cost + Blast Radius)

Goal: Before deploying, understand what it will cost and what could break.

Step 1: Generate a plan file

Plans are framework-specific. Generate one using your IaC tool:

=== “Terragrunt (multi-stack)”

```bash
terragrunt run \
  --working-dir stacks/ \
  --all \
  --out-dir tfplan \
  --json-out-dir tfplan \
  -- plan -lock=false
```

Output: `tfplan/` directory with JSON plans per stack.

=== “Terraform / OpenTofu”

```bash
terraform plan -out=tfplan.binary
terraform show -json tfplan.binary > tfplan.json
```

Output: single `tfplan.json` file.

=== “CloudFormation”

```bash
# Create change set
aws cloudformation create-change-set \
  --stack-name my-stack \
  --template-body file://template.yaml \
  --change-set-name analyze-changes
```

Or analyze directly from template: `--template template.yaml`

=== “CDK v2”

```bash
cdk synth --output cdk.out
```

Analyze the synthesized template: `--template cdk.out/MyStack.template.json`

Step 2: Ask AI to analyze

=== “Via Kiro CLI (AI Agent)”

```bash
kiro-cli chat --agent thoth
```

Then in the chat session:
```
You: "Analyze the cost and blast radius for my planned changes"

AI: [Calls cost_analysis + check_iac with blast-radius]

💰 Cost Impact: +$340/month across 5 stacks
   - EKS node scaling: +$280/mo
   - New RDS read replica: +$60/mo

📊 Blast Radius: HIGH
   - 14 resources affected
   - 3 cross-stack dependencies
   - EKS node group: rolling update (potential 5min downtime)

Recommendation: Deploy foundation layer first, wait 10 min,
then deploy platform layer.
```

=== “Via CLI”

```bash
# Cost analysis
thothctl check iac -type cost-analysis --plan-file tfplan/ --recursive

# Blast radius
thothctl check iac -type blast-radius --plan-file tfplan/ --recursive
```

Try It

# Generate a plan and analyze it
terraform plan -out=tfplan.binary
terraform show -json tfplan.binary > tfplan.json
thothctl check iac -type cost-analysis --plan-file tfplan.json
thothctl check iac -type blast-radius --plan-file tfplan.json

Use Case 5: Detect and Fix Drift

Goal: Find resources that have drifted from your IaC definition and get remediation guidance.

Step 1: Run drift detection

=== “Via Kiro CLI (AI Agent)”

```bash
kiro-cli chat --agent thoth
```

Then in the chat session:
```
You: "Check for drift in my production infrastructure"
```

=== “Via CLI (direct)”

```bash
# Basic drift detection
thothctl check iac -type drift --recursive

# With AI-powered root cause analysis
thothctl check iac -type drift --recursive --ai-provider ollama
```

Step 2: Review findings

🔍 Drift Detection Results:
┌────────────────────────────┬──────────┬─────────────────────────┐
│ Resource                   │ Severity │ What Changed            │
├────────────────────────────┼──────────┼─────────────────────────┤
│ aws_security_group.web     │ HIGH     │ Ingress rule added      │
│ aws_s3_bucket.logs         │ MEDIUM   │ Lifecycle rule removed  │
│ aws_iam_role.lambda        │ LOW      │ Tag modified            │
└────────────────────────────┴──────────┴─────────────────────────┘

🔴 HIGH: aws_security_group.web
   Someone manually added an ingress rule allowing 0.0.0.0/0 on port 22.
   Your IaC restricts SSH to VPN CIDR (10.0.0.0/8) only.
   
   Remediation: Run `terraform apply` to reconcile back to desired state.
   This will REMOVE the dangerous rule.

Step 3: Remediate

# Preview what terraform would change
terraform plan

# Apply to reconcile (removes manual drift)
terraform apply

Try It

# Requires cloud credentials (reads actual state)
thothctl check iac -type drift --recursive

Use Case 6: Generate Documentation

Goal: Auto-generate README, dependency graphs, and architecture diagrams from your IaC code.

Step 1: Generate docs

=== “Via Kiro CLI (AI Agent)”

```bash
kiro-cli chat --agent thoth
```

Then in the chat session:
```
You: "Generate documentation for my infrastructure project"
```

=== “Via CLI (direct)”

```bash
# Generate docs for all stacks
thothctl document iac --recursive

# Terragrunt dependency graph
thothctl document iac --framework terragrunt --graph-type mermaid
```

Step 2: Review output

✅ Generated documentation:
   ├── stacks/foundation/network/vpc/README.md
   ├── stacks/foundation/iam/roles/README.md
   ├── stacks/platform/eks/cluster/README.md
   ├── docs/dependency-graph.mmd (Mermaid)
   └── docs/topology.png (architecture diagram)

Try It

thothctl document iac --recursive

How It All Works Together

The complete AI-DLC flow connects generation, validation, review, and monitoring:

%%{init: {'theme':'base', 'themeVariables': { 'primaryColor':'#e3f2fd','primaryTextColor':'#1565c0','primaryBorderColor':'#1976d2','lineColor':'#42a5f5','secondaryColor':'#fff3e0','tertiaryColor':'#f3e5f5','fontSize':'14px'}}}%%
graph TD
    intent["<b>1. Developer Intent</b><br/><small>'I need a VPC with...'</small>"]:::startNode
    gen["<b>2. Generate IaC</b><br/><small>generate iac → validated code</small>"]:::genNode
    plan["<b>3. Plan & Analyze</b><br/><small>terraform plan → cost + blast radius</small>"]:::planNode
    scan["<b>4. Scan & Review</b><br/><small>scan iac + ai-review → findings</small>"]:::scanNode
    deploy["<b>5. Deploy</b><br/><small>terraform apply</small>"]:::deployNode
    monitor["<b>6. Monitor</b><br/><small>drift detection + dashboard</small>"]:::monitorNode

    intent --> gen
    gen --> plan
    plan --> scan
    scan -->|"✅ Approved"| deploy
    scan -->|"❌ Changes needed"| gen
    deploy --> monitor
    monitor -->|"Drift detected"| scan

    classDef startNode fill:#7c4dff,stroke:#6200ea,stroke-width:2px,color:#fff
    classDef genNode fill:#2196f3,stroke:#1565c0,stroke-width:2px,color:#fff
    classDef planNode fill:#ff9800,stroke:#e65100,stroke-width:2px,color:#fff
    classDef scanNode fill:#e91e63,stroke:#880e4f,stroke-width:2px,color:#fff
    classDef deployNode fill:#4caf50,stroke:#2e7d32,stroke-width:2px,color:#fff
    classDef monitorNode fill:#00bcd4,stroke:#006064,stroke-width:2px,color:#fff

Each step can be run:

Manually via individual CLI commands
Automatically via thothctl workflow devsecops --phase all
Conversationally via AI assistant + MCP

MCP Tools Reference

ThothCTL exposes 26 tools via the Model Context Protocol. Your AI assistant can call any of these:

Category	Tool	What It Does
Generation	`thothctl_generate_iac`	Natural language → governed IaC code
	`thothctl_generate_stacks`	YAML-driven stack generation
Security	`thothctl_scan_iac`	Multi-tool scanning (Checkov, Trivy, KICS, OPA)
	`thothctl_ai_review`	Multi-agent AI security analysis + PR decisions
Analysis	`thothctl_check_iac`	Cost analysis, blast radius, drift, compliance
	`thothctl_cost_analysis`	AWS cost projections
	`thothctl_drift_detection`	Infrastructure drift detection
Workflow	`thothctl_workflow_devsecops`	Run DevSecOps SDLC phases
	`thothctl_workflow_run`	Execute custom YAML DAG workflows
Inventory	`thothctl_inventory_iac`	SBOM, dependencies, version checking
Project	`thothctl_init_project`	Initialize project from scaffold
	`thothctl_init_space`	Create organizational space
	`thothctl_init_env`	Bootstrap development environment
	`thothctl_remove_project`	Remove a project
	`thothctl_remove_space`	Remove a space
	`thothctl_list_projects`	List managed projects
	`thothctl_list_spaces`	List organizational spaces
	`thothctl_list_templates`	List available templates
	`thothctl_project_cleanup`	Clean temporary files
	`thothctl_project_convert`	Convert project ↔ template
Docs	`thothctl_document_iac`	Auto-generate documentation
Ops	`thothctl_check_project`	Project structure validation
	`thothctl_check_environment`	Dev environment verification
	`thothctl_quickstart`	Guided onboarding wizard
	`thothctl_upgrade`	Upgrade ThothCTL
	`thothctl_version`	Version info

Configuration

AI Review Settings (`.thothcf.toml`)

[ai_review]
provider = "ollama"           # ollama | bedrock | openai | azure
model = "llama3.1:8b"         # Model name (provider-specific)

[ai_review.budget]
daily_limit_tokens = 1000000  # Daily token budget
monthly_limit_usd = 50.00     # Monthly cost cap
auto_fallback = true          # Fall back to offline on budget exceeded

[ai_review.safety]
require_human_approval = true # Require human for PR decisions
min_confidence = 0.85         # Minimum confidence for auto-decisions

Environment Variables

Variable	Purpose	Default
`OLLAMA_HOST`	Ollama server URL	`http://localhost:11434`
`AWS_REGION`	AWS region for Bedrock	`us-east-1`
`OPENAI_API_KEY`	OpenAI API key	—
`AZURE_OPENAI_ENDPOINT`	Azure OpenAI endpoint	—
`THOTH_ORG_POLICY`	Git URL for org policies (OPA/Rego)	—

When to Use What

Situation	Approach	Command
“I need new infrastructure”	Generate from intent	`generate iac --intent "..." --provider ollama`
“Is this PR safe to merge?”	AI review + decide	`ai-review decide --pr-number 42`
“Run all checks before deploy”	Workflow pipeline	`workflow devsecops --phase pre-deploy`
“What will this cost?”	Cost analysis	`check iac -type cost-analysis`
“Is production drifting?”	Drift detection	`check iac -type drift`
“Explain findings and fix them”	AI assistant via MCP	Chat: “Review scan results and fix issues”
“Full audit for compliance”	Workflow + scanning	`workflow devsecops --phase all --enforcement hard`

Next Steps

Try generation: thothctl generate iac --intent "S3 bucket with encryption" --project-type terraform --provider ollama --apply
Try AI review: thothctl ai-review analyze -d . -p ollama
Try the pipeline: thothctl workflow devsecops --phase secure
Launch dashboard: thothctl dashboard launch
Explore interactively: Configure MCP and ask your AI assistant “Scan my infrastructure”

DevSecOps SDLC Guide — Detailed phase-by-phase CLI reference with CI/CD examples
MCP Command Reference — Server configuration and troubleshooting
AI Review Command — Multi-agent system details
Generate IaC Command — Intent-to-IaC full reference
Concepts: AI Workflows — How the three AI workflows relate

This site is open source. Improve this page.

thothctl

AI-Powered Development Lifecycle (AI-DLC)

What You Can Do

The AI-DLC Workflow

Prerequisites

1. Install ThothCTL

2. Configure MCP (for AI assistant integration)

3. (Optional) Install a local LLM for AI Review

Verify Setup

Use Case 1: Generate Infrastructure from Natural Language

Step 1: Initialize your project and start the AI agent

Step 2: Describe your intent

Step 2: ThothCTL generates, validates, and self-corrects

Step 3: Review and deploy

Supported Project Types

Try It

Use Case 2: AI-Powered Security Review

Step 1: Run AI review on your code

Step 2: Four AI agents analyze in parallel

Step 3: Make a PR decision (optional)

Available Providers

Try It

Use Case 3: Run DevSecOps Pipeline via AI

Option A: Via Kiro CLI (conversational)

Option B: Via CLI (for CI/CD)

Option C: Custom YAML workflow

Available Phases

Try It

Use Case 4: Analyze Planned Changes (Cost + Blast Radius)

Step 1: Generate a plan file

Step 2: Ask AI to analyze

Try It

Use Case 5: Detect and Fix Drift

Step 1: Run drift detection

Step 2: Review findings

Step 3: Remediate

Try It

Use Case 6: Generate Documentation

Step 1: Generate docs

Step 2: Review output

Try It

How It All Works Together

MCP Tools Reference

Configuration

AI Review Settings (.thothcf.toml)

Environment Variables

When to Use What

Next Steps

Related Docs

AI Review Settings (`.thothcf.toml`)