# Eng Security Safety > Apply proactive threat modeling, least-privilege design, and safety guardrails before delivering any code or infrastructure change. Use when this capability is needed. - Skill: `tomevault-io/eng-security-safety` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/eng-security-safety` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/eng-security-safety/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Security - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/eng-security-safety --- # Security and Safety Mindset ## Intent - Treat every change as a potential attack surface or failure amplifier. - Ensure data classification, secret handling, and permission scopes stay compliant. - Bake safety checks (rate limits, input validation, monitoring) into the design, not after. ## Baseline Checklist 1. **Threat model quickly**: Who could abuse this surface? What capabilities do they need? What happens if they succeed? 2. **Data stewardship**: Classify data touched (PII, payments, assets) and enforce encryption, retention, and locality rules. 3. **Access + identity**: Validate authn/authz paths, key rotation, wallet signatures, and privilege escalation barriers. 4. **Dependency hygiene**: Pin versions, verify licenses, review changelogs, and prefer audited libraries/contracts. 5. **Secrets + config**: Never log secrets; store them in the project’s approved secret manager. Guard env var usage. ## Workflow 1. Enumerate entry points (mobile UI, API, smart contract, admin tools) and list unchecked inputs. 2. Define validation layers: schema-level, business-level, and environment-level (e.g., chain ID, platform version). 3. Ensure every state change is reversible or compensatable (feature flags, contract pausing, migration guards). 4. Instrument detection: structured logs, metrics, or on-chain events that can surface abuse or regressions fast. 5. Document explicit “never do” actions (e.g., disable signature checks, bypass paywalls) inside the PR/issue notes. ## Verification - Run the project’s security/static analysis tooling (linters, contract analyzers, mobile scanners) and fix findings. - Peer review the threat model summary; confirm secrets and keys are absent from diffs/logs. - Validate abuse cases end-to-end (invalid payloads, replayed signatures, abusive traffic) before shipping. --- > Converted and distributed by [TomeVault](https://tomevault.io/claim/tjboudreaux) — claim your Tome and manage your conversions.