Skip to content

Security: EM1805/causal-safety-engine

Security

docs/SECURITY.md

Causal Safety Engine— Security and Privacy

Causal Safety Engine is designed as a local-first system.

Network

  • No mandatory network calls
  • No telemetry collection required for operation
  • All inputs and outputs remain on the executing machine

Local orchestration

Run scripts and smoke tests may invoke local subprocesses to orchestrate pipeline steps. These subprocesses do not imply network activity.

Artifacts

  • Artifacts are written to local disk only (under CANONICAL/**/out/)
  • Users are responsible for storage, access control, and any organizational compliance requirements

Security Policy

Supported Versions

This project is provided as a local-first analytical tool. No network-exposed services are included.

Threat Model

  • CLI-only execution
  • Local filesystem access only
  • No remote inputs, no authentication layer
  • No multi-tenant or privileged execution

Reporting a Vulnerability

If you believe you have found a security issue, please contact: security@ (or GitHub private disclosure)

Known Limitations

Some CLI arguments are used as filesystem paths. This is a conscious design choice for local-first workflows.

Threat Model Summary

Vector Status
Remote code execution Not applicable
Network attack surface None
Privilege escalation Not applicable
Local file overwrite Accepted risk (CLI user-controlled)
Data exfiltration Not applicable

Accepted Security Risks

  • Path traversal warnings reported by static analysis tools are accepted for CLI-controlled output paths.
  • The tool assumes a trusted local execution environment.

There aren't any published security advisories