Security Policies & ZIP Taint
Relevant source files
The following files were used as context for generating this wiki page:
The Lou Gehrig Fan Club (LGFC) project maintains a rigorous security posture designed to prevent unauthorized access, configuration drift, and the introduction of malicious or opaque artifacts. The primary security mechanisms include strict Git authentication protocols, automated configuration snapshots for Cloudflare Pages, and a "no-artifact" policy that specifically blocks ZIP files to prevent repository "taint."
These policies ensure that the repository remains a transparent, source-available environment. All contributors, including human developers and AI Agents, are subject to these safety gates to maintain the integrity of the LGFC website and FanClub area.
Artifact Governance & ZIP Taint Prevention
To maintain repository transparency and security, the project enforces a strict policy against committing binary artifacts, specifically ZIP files. This "ZIP Taint" prevention strategy ensures that all code remains reviewable and that no hidden logic or dependencies are introduced through compressed archives.
Artifact Restrictions
The repository explicitly forbids the inclusion of the following items:
- ZIP Files: Any form of compressed archive.
- Build Output: Artifacts generated during the build process.
- Temporary Files: Screenshots, local-only logs, or intermediate artifacts.
Sources: governance/ai/AGENT-GOVERNANCE.md:46-49, PROMPTS/Codex-Rules.md:14-14, README.md:47-49
Governance Enforcement Flow
The following diagram illustrates how artifact governance is applied during the contribution lifecycle:
The repository utilizes automated checks to ensure that Agents modify only allowlisted files and do not introduce forbidden artifacts.
Sources: PROMPTS/CodEX-Rules.md:7-14, governance/ai/AGENT-GOVERNANCE.md:38-42
Authentication & Access Control
Security is anchored in strict identity verification for both local development and GitHub Codespaces. The project has moved away from password-based authentication in favor of token-based and SSH-based systems.
Git Authentication Methods
| Method | Usage Scenario | Requirements |
|---|---|---|
| Personal Access Token (PAT) | Codespaces / Local HTTPS | Requires repo and workflow scopes. |
| SSH Keys | Local Development | Requires ED25519 or RSA keys linked to GitHub. |
GitHub CLI (gh) | Codespaces Interactive | Pre-installed; uses gh auth login flow. |
Sources: CONTRIBUTING.md:45-70, CONTRIBUTING.md:103-120
Codespaces Security
In GitHub Codespaces, the implicit token often lacks the necessary permissions for Git operations. Contributors must configure a PAT or use the GitHub CLI to prevent authentication failures. The repository guidelines warn never to commit these tokens to the repository.
Sources: CONTRIBUTING.md:75-84, CONTRIBUTING.md:122-122
Cloudflare Pages Security Snapshots
To prevent configuration drift and enable secure recovery, the project utilizes an automated snapshot system for Cloudflare Pages metadata.
Data Redaction & Security
The snapshot process is designed to be "read-only" and secure:
- No Secrets: API tokens and secret values are never exported or written to disk.
- Environment Variables: Only the names of environment variables are captured, not their values.
- Redaction: The snapshot script validates all responses and redacts sensitive data before saving JSON files.
Sources: snapshots/cloudflare/README.md:39-50, scripts/cf_pages_snapshot.sh:119-122
Snapshot Lifecycle
The snapshot automation runs daily at 07:00 UTC and on-demand, ensuring a verifiable history of the production state without compromising credentials.
Sources: scripts/cf_pages_snapshot.sh:37-65, snapshots/cloudflare/README.md:52-57
Agent Governance & Safety Gates
AI Agents operate under a strict execution model to prevent accidental security breaches or scope expansion.
Operational Guardrails
- Scope Limitation: Agents must work from one source issue and a single thread.
- Verification Evidence: Every agent handoff must include verification commands run and a result summary.
- Forbidden Modifications: Agents are prohibited from modifying unrelated files or introducing unapproved architecture changes.
Sources: governance/ai/AGENT-GOVERNANCE.md:27-36, Agent.md:44-48
Handoff Requirements
| Requirement | Description |
|---|---|
| Source Issue | Mapping the work to a specific requirement. |
| Files Changed | Explicit list of touched paths. |
| Verification | Evidence of testing (e.g., npm run lint). |
| Limitations | Disclosure of any risks or incomplete items. |
Sources: governance/ai/AGENT-GOVERNANCE.md:51-58, ops/ai/CROSS-AGENT-OPERATING-RULES.md:41-48
Summary
The Security Policies of the LGFC project focus on transparency and artifact integrity. By enforcing a strict "No ZIP" policy, the project eliminates the risk of binary taint. Automated Cloudflare snapshots provide a secure, redacted audit trail of the infrastructure, while rigid Agent governance ensures that AI-driven contributions remain within defined, safe boundaries. All security operations prioritize token-based authentication and automated verification to maintain a production-ready environment.