---
name: "mechanical-strength-review"
description: "Perform preliminary, vendor-neutral strength and stiffness screening for small mechanical parts using explicit load cases, free-body diagrams, first-order mechanics, applicability checks, and evidence-aware margins. Use when asking whether a bracket, beam, shaft, pin, boss, snap arm, or similar part may carry a stated load; when estimating stress, strain, deflection, bearing, torsion, thermal expansion, or Euler buckling; or when deciding whether hand calculations, FEM, coupons, or a qualified engineer are required."
metadata:
  dev.v-san.skills: |-
    origin: "original (personal skill)"
    version: "0.1"
    changeDate: "2026-08-01"
    requires:
      dependencies:
        - "Python 3.10\u002B for bundled calculators"
    sources:
      - "https://web.mit.edu/course/3/3.11/www/module_list.html https://ocw.mit.edu/courses/2-001-mechanics-materials-i-fall-2006/pages/lecture-notes/ https://standards.nasa.gov/node/263\n"
---

# Mechanical Strength Review

Produce a preliminary engineering check, not a certification or an unconditional safety verdict.

## Workflow

1. Define the consequence of failure. Escalate pressure, lifting people or loads over people, high stored energy, regulated products, and other serious hazards to a qualified engineer.
2. Record units, service environment, load cases, load directions, duration, cycles, impact or misuse factors, supports, and interfaces. Draw or describe a free-body diagram and verify equilibrium.
3. Enumerate plausible failure modes before calculating: yielding or rupture, excessive deflection, bearing, shear-out, pullout, local buckling, global buckling, creep, fatigue, joint slip, or loss of retention.
4. Select the smallest defensible model from [first-order-models.md](references/first-order-models.md). State every applicability condition and violated assumption.
5. Run `scripts/mechanics_check.py` for supported calculations. Keep its explicit N-mm-MPa unit system in the report.
6. Compare against an allowable only when its source, material condition, direction, temperature, and process are recorded. Never substitute a generic material-family value silently.
7. Choose the next evidence level: geometry correction, refined analytical model, FEM verification, process-matched coupon, assembly test, or engineer review.
8. Return the contract in [evidence-and-report.md](references/evidence-and-report.md). Keep untested claims visible.

## Claim discipline

- Treat a hand calculation as screening within its stated idealization.
- Do not apply circular-shaft torsion formulas to arbitrary sections.
- Do not treat von Mises stress as a universal failure criterion.
- Do not report a margin when the allowable lacks provenance.
- Do not turn missing loads, material evidence, or boundary conditions into a pass.
- Separate model verification from validation against the manufactured part.

## Resources

- [first-order-models.md](references/first-order-models.md) — equations, units, and stop conditions.
- [evidence-and-report.md](references/evidence-and-report.md) — evidence ladder and report contract.
- `scripts/mechanics_check.py` — dependency-free JSON calculator for supported screening models.