Security Development Lifecycle (SDL)
IEC 62443-4-1 SM-1 Compliance
1. Purpose
This document defines the formal Security Development Lifecycle (SDL) for the
RUNE platform. It establishes the security management system required by
IEC 62443-4-1 SM-1, ensuring that security is integrated into every phase of
the development process.
2. Scope
This SDL applies to all RUNE repositories:
| Repository |
Language |
Description |
rune |
Python |
Core CLI and orchestration engine |
rune-operator |
Go |
Kubernetes operator |
rune-ui |
HTMX |
Web interface |
rune-charts |
Helm |
Deployment charts |
rune-docs |
MkDocs |
Documentation (this repo) |
rune-audit |
Mixed |
Audit tooling |
3. Security Objectives
- Confidentiality -- Secrets (API keys, tokens) never persist in code or
configuration files; environment variables only.
- Integrity -- All releases carry SLSA L3 signed provenance. Container
images are signed with cosign (see IMAGE_SIGNING.md).
- Availability -- Fail-closed cost gates and resource limits prevent
runaway provisioning.
- Traceability -- Every change is traceable from issue to merge via branch
protection and mandatory CI gates.
4. Roles and Responsibilities
| Role |
Responsibility |
Project Owner (lpasquali) |
Final authority on risk acceptance; reviews all security-relevant PRs |
| Security Lead |
Owns threat model, risk register (RISK_REGISTER.md), and incident response (INCIDENT_RESPONSE.md) |
| Developer / Agent |
Follows SDL phases; runs audit checks per SYSTEM_PROMPT.md triggers |
| CI Pipeline |
Enforces automated gates (SAST, SCA, SBOM, license compliance, coverage) |
5. SDL Phases
flowchart LR
A[Requirements] --> B[Design]
B --> C[Implementation]
C --> D[Verification]
D --> E[Release]
E --> F[Response]
F -->|Lessons learned| A
5.1 Requirements
- Threat modeling using STRIDE methodology (see RISK_ASSESSMENT.md).
- Security requirements derived from IEC 62443-4-1 and SLSA L3.
- Privacy and data-handling requirements documented per feature.
5.2 Design
- Architecture Decision Records (ADRs) for security-relevant changes.
- DriverTransport protocol security review for new agent integrations.
- API versioning policy (additive changes preferred; no silent breaking changes).
5.3 Implementation
- Secrets in environment variables only -- never in
rune.yaml or source.
- All dependencies scanned before introduction (
pip-audit, grype, safety).
- License gate blocks copyleft licenses (AGPL, GPL-2.0, GPL-3.0) at CI.
- 97% code coverage floor enforced by CI.
5.4 Verification
- Static Analysis (SAST): Trivy filesystem scan on every PR.
- Software Composition Analysis (SCA): SBOM generation via Syft; vulnerability scanning via Grype and Trivy.
- Secret Scanning: Gitleaks on every push.
- Penetration Testing: Quarterly and pre-release (see PENTEST.md).
- Fuzz Testing: Integrated into CI for protocol and API parsers (see FUZZ_TESTING.md).
- CVE Policy: Fixable vulnerabilities above CVSS 8.8 block merge. Unfixable above threshold require fork-and-patch.
5.5 Release
- Signed provenance attestation (SLSA L3) on all artifacts.
- Container image signing with cosign via Fulcio/Rekor (see IMAGE_SIGNING.md).
- SBOM published as release artifact.
- VEX statements maintained for known exceptions (see
delivery/VEX.md).
5.6 Response
- Incident response plan with severity classification and SLAs (see INCIDENT_RESPONSE.md).
- Post-incident reviews feed back into Requirements phase.
- Vulnerability disclosure process via GitHub Security Advisories.
6. Review Cadence
| Activity |
Frequency |
| Threat model review |
Quarterly |
| Risk register review |
Quarterly |
| Penetration test |
Quarterly + pre-release |
| Full audit (legal + cyber) |
Milestone exit / quarterly |
| SDL document review |
Annually or after major incident |
| Dependency audit |
Every PR (automated) |
7. Training Requirements
| Audience |
Training |
Frequency |
| All developers |
Secure coding practices (OWASP Top 10) |
Annually |
| All developers |
RUNE SDL process and CI gate operation |
Onboarding + annually |
| Security Lead |
IEC 62443-4-1 requirements and audit procedures |
Annually |
| AI Agents |
SYSTEM_PROMPT.md consumption at boot (automatic) |
Every session |
8. Metrics
- Mean time to remediate critical vulnerabilities (target: < 48 hours).
- Percentage of PRs passing all security gates on first submission.
- Coverage floor adherence (97% minimum).
- Number of risk acceptances in the risk register.
9. References