Igor v0 is experimental research software currently in Phase 2 (Survival) development. It is not production-ready and has known security limitations by design.

Current Status: Early development, active testing, not suitable for production use.

Only the main branch is actively maintained. No stable releases have been tagged yet.

Igor includes security-sensitive components:

• Agent isolation via wazero runtime
• Memory limits (64MB per agent)
• Capability restrictions (no filesystem, no network)
• Execution timeouts (100ms per tick)

• Agent package transfer over libp2p streams
• Checkpoint serialization and transmission
• Confirmation handshake protocol

• State serialization and storage
• Budget metadata encoding
• Atomic write operations

• Execution time measurement
• Budget calculation and enforcement
• Cost deduction mechanisms

Igor v0 has intentional security limitations:

Budget Security:

• Budget accounting is trusted, not cryptographically verified
• Nodes can lie about execution time
• No payment receipts or proof of work
• No dispute resolution mechanism

Checkpoint Security:

• Checkpoints stored in plaintext
• No encryption of agent state
• No integrity verification
• No authentication of checkpoint origin

Network Security:

• No peer authentication beyond libp2p peer ID
• No authorization for migration requests
• No rate limiting or DoS protection
• Relies on trusted network environment

Identity Security:

• Agents lack cryptographic identity in v0
• No signing capability
• No access control
• Identity not carried in migrations

These limitations are documented in docs/runtime/SECURITY_MODEL.md.

Do not deploy Igor v0 on public networks or with sensitive data.

If you discover a security vulnerability in Igor, please report it responsibly.

1. Go to https://github.com/simonovic86/igor/security/advisories
2. Click "Report a vulnerability"
3. Provide detailed description
4. Allow time for assessment and patch development

Email security issues to: security@igor-project.org (placeholder - use GitHub issues for now)

• Description of vulnerability
• Steps to reproduce
• Potential impact
• Suggested fix (if applicable)
• Your contact information (for coordination)

Critical:

• WASM sandbox escapes
• Arbitrary code execution on host
• Memory corruption or buffer overflows
• Resource exhaustion attacks that crash nodes
• Migration protocol attacks that create duplicate agents
• Checkpoint corruption that breaks recovery

Important:

• Denial of service vulnerabilities
• Information disclosure
• Privilege escalation within sandbox
• Budget manipulation exploits
• Migration abuse scenarios

Lower Priority (but still valuable):

• Logging security issues
• Configuration weaknesses
• Documentation improvements

We follow coordinated disclosure:

1. Report received - We acknowledge within 48 hours
2. Assessment - We evaluate severity and impact
3. Fix development - We develop and test a patch
4. Disclosure timeline - We agree on public disclosure date
5. Patch release - We release fix to main branch
6. Public disclosure - We publish security advisory

Typical timeline: 30-90 days from report to disclosure, depending on severity.

• Malicious agents escaping WASM sandbox
• Resource exhaustion by single agent (memory, CPU)
• State corruption through atomic checkpoint writes

• Malicious nodes lying about metering or stealing state
• Network attacks beyond basic libp2p security
• Cryptographic vulnerabilities (minimal crypto in v0)
• Economic attacks (no payment verification)
• Data privacy (checkpoints in plaintext)

Igor v0 is suitable only for:

• Development and testing
• Research environments
• Trusted network deployments
• Non-sensitive agent workloads

Future phases may address current limitations:

Phase 3 (Autonomy):

• Agent manifest validation
• Capability enforcement
• Basic integrity checks

Phase 4 (Economics):

• Cryptographic receipts
• Payment verification
• Fraud detection

Phase 5 (Hardening):

• State encryption
• Checkpoint signing
• Advanced sandbox hardening
• Multi-party verification

No timeline or commitment. Listed for context only.

• docs/runtime/SECURITY_MODEL.md - Detailed threat model
• docs/enforcement/RUNTIME_ENFORCEMENT_INVARIANTS.md - System guarantees
• PROJECT_CONTEXT.md - Design philosophy

For non-sensitive questions about Igor security:

• Open a GitHub issue
• Use "security" label
• Public discussion welcome

For sensitive vulnerability reports:

• Use GitHub Security Advisory
• Private disclosure until patched