adryserage-aetheris 3.0.0

Aetheris - Multi-agent AI code analysis system with PR review, multi-provider consensus (Gemini/Claude/OpenAI), security analysis, and automatic bug fixing via Claude Code

Aetheris — AI Code Review That Actually Works

10 AI agents. Automated vulnerability detection. Minutes instead of hours.

View Full Download History | Detailed Statistics

Quick Start | Demo | Architecture | Documentation

---

Help Shape Aetheris

Should Aetheris become Open Source? Cast your vote

---

Stop Shipping Bugs. Start Shipping Confidence.

Your code reviews are broken. Senior devs spend hours reviewing PRs. Junior devs miss critical bugs. Security vulnerabilities slip into production. Sound familiar?

Aetheris fixes this. It's like having a senior developer, security expert, and architect review every commit — in minutes, not hours.

Your Pain Point	Aetheris Solution
"Code review takes forever"	Fast automated scans
"We keep shipping the same bugs"	Consistent, thorough reviews 24/7
"Security audits are expensive"	Built-in CVE + OWASP scanning
"Junior devs miss edge cases"	AI catches what humans miss
"We're locked into [expensive tool]"	Pay-per-use, switch providers anytime

pip install adryserage-aetheris && aetheris analysis

That's it. No account creation. No credit card. No 30-minute setup call.

---

Features

• Interactive CLI (v3.0) — REPL interface with aetheris --cli for conversational code analysis
• Skills System (v3.0) — AI agents can read/write files, execute commands, and search code
• MCP Integration (v3.0) — Connect to Model Context Protocol servers for extended capabilities
• Custom Skills (v3.0) — Define your own skills via YAML configuration
• Headless Mode (v2.8) — Non-interactive execution for CI/CD with aetheris run <task-file>
• Deterministic Execution — Reproducible results with --seed and --no-timestamps options
• Security Profiles — Control operations with --profile safe|ci|dev|unrestricted
• Configurable Timeout — Set analysis duration with --timeout 30m, --timeout 2h, or --timeout 0 (unlimited)
• Heartbeat Monitor — Progress indicator every 10 minutes for long-running analyses
• Incremental Reanalysis — Use --reanalyze / -r to analyze only files changed since last run
• Smart Change Detection — Auto-detects modifications via git diff or file timestamps
• Cost-Efficient Updates — Skip unchanged files, preserve previous results, reduce API calls
• Graceful Interruption — Partial results saved when timeout is reached
• Multi-Provider Support — Switch between Gemini, OpenAI, and Claude with --provider
• Auto-Fix Integration — Fix issues automatically with Claude Code via aetheris fix --auto
• GitHub Integration — Create issues, review PRs, generate fix PRs directly
• Smart Caching — 60-80% cost reduction on repeat analyses
• Artifact Cleanup — Remove development artifacts before packaging with aetheris cleanup
• Secret Scanner — Detect exposed credentials with confidence levels (high/medium/low)
• Package Validation — Pre-publication security checks with aetheris validate-package
• Multi-Language Support — CLI available in English and French with --lang flag or AETHERIS_LANG env var

Traditional Tools	Aetheris
Thousands of noisy warnings	Actionable insights only
Pattern matching (shallow)	Context-aware AI (deep)
Security OR quality OR architecture	All 10 agents working together
Suggestions you ignore	Auto-fix that actually works

Pricing That Makes Sense

No subscriptions. No per-seat licensing. Just pay for what you use.

What You Get	Enterprise SaaS	Aetheris
Monthly cost (10 devs)	$200-500/mo	$5-20/mo
Your code stays...	On their servers	With your AI provider
Vendor lock-in	Yes	Switch providers anytime
Auto-fix capability	Suggestions only	Actual code changes

Real Cost Example (1,000 files)

Provider	Economy Model	With 70% Cache
Gemini Flash	~$3-10	~$1-3
GPT-4o-mini	~$5-15	~$2-5
Claude Haiku	~$10-20	~$3-6

Smart caching skips unchanged files. Expect 60-80% cost reduction on repeat analyses.

Pricing Note: AI API pricing changes frequently. Check current rates: Gemini | OpenAI | Anthropic

---

How It Works

Architecture

Aetheris coordinates 10 specialized AI agents through a central orchestrator. Each agent analyzes a specific domain, and the QA Synthesis agent consolidates findings into prioritized, actionable reports.

10 Specialized AI Agents

Core Agents (6)

Agent	Responsibility
Security	SQL injection, XSS, secrets in code, OWASP Top 10, weak crypto
Architecture	Coupling issues, circular dependencies, anti-patterns, layering violations
Code Quality	Dead code, complexity hotspots, edge cases, maintainability issues
Metrics	Cyclomatic complexity, duplication, maintainability index, LOC stats
Dependency	CVE scanning via OSV database, outdated packages, license compliance
QA Synthesis	Prioritizes findings, eliminates false positives, generates final report

Extended Agents (4) — v2.10+

Agent	Responsibility
Type Safety	any usage, unsafe casts, type ignores, null safety issues
Performance	O(n²) complexity, N+1 queries, memory leaks, blocking async
API Contract	REST violations, HTTP method issues, breaking changes, undocumented endpoints
Data Privacy	PII logging, missing encryption, GDPR/HIPAA/PCI-DSS/CCPA compliance

Integration Points

Integration	How It Works
AI Providers	Gemini (default), OpenAI, Claude — switch with --provider flag
GitHub	Create issues from findings, review PRs, generate fix PRs
CI/CD	GitHub Actions template included, GitLab CI, Jenkins compatible
Output Formats	Markdown reports, JSON metrics, console summaries

---

Features

• Headless Mode (v2.8) — Non-interactive execution for CI/CD with aetheris run <task-file>
• Deterministic Execution — Reproducible results with --seed and --no-timestamps options
• Security Profiles — Control operations with --profile safe|ci|dev|unrestricted
• Configurable Timeout — Set analysis duration with --timeout 30m, --timeout 2h, or --timeout 0 (unlimited)
• Heartbeat Monitor — Progress indicator every 10 minutes for long-running analyses
• Incremental Reanalysis — Use --reanalyze / -r to analyze only files changed since last run
• Smart Change Detection — Auto-detects modifications via git diff or file timestamps
• Cost-Efficient Updates — Skip unchanged files, preserve previous results, reduce API calls
• Graceful Interruption — Partial results saved when timeout is reached
• Multi-Provider Support — Switch between Gemini, OpenAI, and Claude with --provider
• Auto-Fix Integration — Fix issues automatically with Claude Code via aetheris fix --auto
• GitHub Integration — Create issues, review PRs, generate fix PRs directly
• Smart Caching — 60-80% cost reduction on repeat analyses
• Artifact Cleanup — Remove development artifacts before packaging with aetheris cleanup
• Secret Scanner — Detect exposed credentials with confidence levels (high/medium/low)
• Package Validation — Pre-publication security checks with aetheris validate-package
• Multi-Language Support — CLI available in English and French with --lang flag or AETHERIS_LANG env var

# Full analysis
aetheris analysis

# Incremental reanalysis (changed files only)
aetheris analysis --reanalyze

# With timeout for large codebases
aetheris analysis -r --timeout 30m

# Headless mode for CI/CD (v2.8)
aetheris run task.json --profile ci

# Auto-fix issues
aetheris fix --analyze-first --auto

---

See It In Action

# Analyze your entire codebase
$ aetheris analysis

  Scanning 127 files...
  Running 10 AI agents...

  CRITICAL: SQL injection in user_service.py:45
  HIGH: Hardcoded API key in config.py:12
  HIGH: N+1 query pattern in orders.py:78-92
  MEDIUM: Circular dependency: auth -> users -> auth

  23 issues found | 4 critical | 8 high | 11 medium
  Report: docs/analyses/quality_assurance_report.md

# Auto-fix with Claude Code
$ aetheris fix --analyze-first --auto
  Fixed 19/23 issues automatically

---

Get Started in 30 Seconds

# 1. Install
pip install adryserage-aetheris

# 2. Configure (free Gemini API key: https://ai.google.dev)
echo "GEMINI_API_KEY=your_key" > .env
echo ".env" >> .gitignore  # IMPORTANT: Never commit API keys

# 3. Analyze
aetheris analysis

---

Complete Command Suite

Command	Purpose	Example
aetheris analysis	Deep codebase scan	aetheris analysis --reanalyze --timeout 30m
aetheris run	Headless execution for CI/CD	aetheris run task.json --profile ci
aetheris stats	View statistics & ROI dashboard	aetheris stats --period week --export json
aetheris fix	Auto-fix with Claude Code	aetheris fix --analyze-first --auto
aetheris pr	GitHub PR reviews	aetheris pr --url <pr-url>
aetheris issues	Create GitHub issues	aetheris issues --labels security
aetheris user-story	Flow analysis	aetheris user-story --auto
aetheris pr-gen	Generate fix PRs	aetheris pr-gen --link-issues
aetheris cleanup	Remove artifacts before packaging	aetheris cleanup --dry-run
aetheris validate-package	Pre-publication security check	aetheris validate-package --fix

Full command documentation

---

Works With Your Stack

Python | TypeScript | JavaScript | Go | Rust | Java | Kotlin | Swift | Dart/Flutter | C/C++ | C# | PHP | Ruby

AI Providers: Gemini | OpenAI | Claude (Anthropic)

CI/CD: GitHub Actions template included | GitLab CI | Jenkins

---

Roadmap

Version	Status	Highlights
v2.8	✅ Done	Headless mode for CI/CD automation
v2.9	✅ Done	Persistent config (config.json)
v2.10	✅ Done	4 new agents (performance, API, type safety, data privacy)
v2.11	✅ Done	Agent selection (--agents, --preset, --category)
v2.12	✅ Done	Safety profiles (--profile safe|ci|yolo)
v2.13	✅ Done	CI/CD templates (GitHub Actions workflow)
v3.0	✅ Done	Interactive CLI, Skills system, MCP integration

---

Documentation

Full Documentation — Complete guides, tutorials, and API reference

Getting Started	Reference	Advanced
Quick Start Guide	CLI Commands	GitHub Actions
Installation	AI Providers	Extending Agents
Configuration	Analysis Agents	Troubleshooting

Local docs: docs/ — Markdown source files

---

Important Safety Information

Safety Profiles (v2.12)

Aetheris provides three safety profiles to control execution boundaries:

Profile	Network	Diff Limit	Repair Loops	Best For
safe (default)	AI only	500 lines	3	Local development
ci	Full	1000 lines	5	CI/CD automation
yolo	Full	Unlimited	10	Power users

# Safe mode (default) - recommended
aetheris analysis --profile safe

# CI mode - for automated pipelines
aetheris analysis --profile ci

# YOLO mode - requires confirmation
aetheris analysis --profile yolo --yes

Safe mode restrictions:

• Blocks auxiliary network calls (web search disabled)
• Validates commands against allowlist
• Limits diff size to 500 lines
• Limits repair loops to 3 iterations

CI mode features:

• Fail-fast on errors (non-zero exit)
• No interactive prompts
• Audit logging to aetheris-audit.jsonl
• Machine-readable output

Data Privacy Notice

Aetheris sends your source code to third-party AI providers (Google Gemini, OpenAI, or Anthropic) for analysis. Before using on proprietary codebases:

1. Review your organization's data policies — Some companies prohibit sending source code to external services
2. Use Enterprise/Zero Data Retention plans:
   • Google Vertex AI — Data not used for training
   • OpenAI Enterprise — Zero data retention
   • Anthropic API — Data not used for training by default
3. Check data residency requirements — For regulated industries (healthcare, finance)

Your API keys are never shared — Aetheris connects directly to your chosen provider.

Auto-Fix Safety Guidelines

The --auto flag applies AI-generated code changes without human review. Risks include:

Risk	Description
Hallucinations	LLMs may generate plausible but incorrect code
Security vulnerabilities	Fixes may introduce new attack vectors
Business logic removal	AI may delete critical code paths

Safe workflow:

aetheris fix --analyze-first --auto  # Run fixes
git diff                              # Review ALL changes (mandatory)
pytest                                # Run tests
git add -p                            # Stage interactively

---

Ready to Ship Better Code?

pip install adryserage-aetheris && aetheris analysis

Install from PyPI | View on GitHub | Get Free Gemini API Key

---

_{Built with care by Adryan Serage}