compiler-optimizer

Purpose

Identify misconfigured or absent Solidity compiler optimizer settings in foundry.toml and pragma declarations. Mismatched optimizer_runs, missing via_ir, and outdated compiler versions are free gas savings requiring only configuration changes.

When to Use

• Writing or reviewing foundry.toml
• Starting a new Foundry project
• Pre-deployment audit of build configuration
• After reviewing pragma versions in .sol files
• When forge test --gas-report shows unexpectedly high gas costs

When NOT to Use

• Non-Foundry projects (Hardhat, Truffle): configuration format differs
• When optimizer is intentionally disabled for faster test iteration (flag for production build profile, not dev profile)
• When the project has recently passed a security audit on a specific version and re-audit is not feasible

Platform Detection

Trigger fires when:

• foundry.toml is present in the project root, or
• A pragma solidity declaration is being reviewed in a .sol file

Read foundry.toml with the Read tool. Use Bash to run forge build --sizes or forge test --gas-report when measurements are needed to quantify savings.

Quick Reference

optimizer_runs by contract type

Contract type	Recommended optimizer_runs	Reason
DeFi hot path (AMM, lending, swap router)	1_000_000	Runtime savings outweigh bytecode size
Factory / deployer contract	1	Minimize deployment cost; called rarely
General purpose (default)	200	Balanced — Foundry default
Library deployed once, called by many	1_000_000	Amortized deployment cost is negligible

All profiles should have optimizer = true. Never deploy with optimizer disabled.

via_ir

Condition	Recommendation
Complex DeFi contract (tight loops, heavy structs, multi-call)	via_ir = true
Simple contract (< 5 functions, minimal logic)	optional
Rapid development iteration	via_ir = false (5–50× slower to compile)
Production build profile	via_ir = true always recommended

Solidity version

Version	Key gain
0.8.22+	Auto-unchecked simple loop counters (~30 gas/iter saved automatically)
0.8.24+	Native tstore/tload syntax for transient storage (SL-005, SL-006)
latest stable	Look up the current release at https://github.com/ethereum/solidity/releases before recommending a specific version — do not assume from memory

Always pin to an exact version (pragma solidity X.Y.Z;), not a floating range (^0.8.0). When recommending a version upgrade, check the Solidity release page for the current stable release and any breaking changes since the project's current pragma.

Workflow

1. Read foundry.toml and check optimizer settings. Confirm optimizer = true. Check optimizer_runs value. Identify whether the project is a DeFi protocol (high-frequency), a deployment-only factory, or a general-purpose contract. Compare the current optimizer_runs to the table above. Check whether via_ir is set, especially for complex contracts with nested internal calls or heavy loop bodies.

2. Check pragma version in .sol files. Scan key contracts for pragma solidity declarations. Identify the current version. If below 0.8.22, the auto-unchecked loop optimization is absent. If below 0.8.24, transient storage requires inline assembly. If below 0.8.28, other incremental optimizer improvements are missing. Note: version upgrades should be validated with forge test to confirm no regressions.

3. Match optimizer_runs to expected call frequency; recommend via_ir for complex contracts. For each finding, quantify the expected impact using forge test --gas-report before and after. Document the change in foundry.toml and the rationale. For via_ir, note the compile-time tradeoff and recommend enabling only in the production profile.

Supporting Docs

Only read these files when explicitly needed — do not load all three by default:

File	Read only when…
resources/PATTERNS.md	You need the break-even calculation formula or full multi-profile foundry.toml examples beyond what's shown above
resources/CHECKLIST.md	Producing a formal /decipher-gas-optimizoor:analyze report and confirming all compiler settings were reviewed
resources/EXAMPLE_FINDING.md	Generating a report and needing the exact output format for an optimizer configuration finding
docs/evm-gas-reference.md	You need EIP references or baseline opcode costs to contextualize compiler optimization impact

Output Format

Report each finding with: pattern ID, config key, current value, recommended value, and estimated gas impact.

Example finding (CO-001):

CO-001 | foundry.toml | optimizer_runs = 200
  Severity : medium
  Context  : AMM swap router called ~1M times/year

  Current foundry.toml:
    [profile.default]
    optimizer = true
    optimizer_runs = 200

  Recommended:
    [profile.default]
    optimizer = true
    optimizer_runs = 1_000_000

    [profile.deploy]
    optimizer = true
    optimizer_runs = 1           # for one-shot factory deployment scripts

  Rationale: At 1M calls/year, a 5% per-call runtime saving at optimizer_runs
  = 1_000_000 vs 200 is: 1,000,000 calls × 50 gas saved × $0.0000001/gas
  ≈ meaningful cumulative savings. Deployment bytecode may grow ~5–10% but
  is a one-time cost.

  Measure:
    forge build --sizes                  # bytecode size delta
    forge test --gas-report              # per-function gas delta

Example finding (CO-002):

CO-002 | foundry.toml | via_ir not set (defaults to false)
  Severity : medium
  Context  : DeFi lending protocol with complex internal call chains

  Current foundry.toml:
    [profile.default]
    optimizer = true
    optimizer_runs = 200

  Recommended:
    [profile.default]
    optimizer = true
    optimizer_runs = 200
    via_ir = true

  Rationale: via_ir lowers code to Yul before optimization, enabling
  cross-function optimizations unavailable to the classic pipeline. For
  contracts with complex internal calls and tight loops, typical savings
  are 5–20% in bytecode size or runtime gas.

  Warning: via_ir increases compile time 5–50×. Enable in production
  profile; keep disabled in dev profile for fast iteration.

  Measure:
    forge build --sizes                  # compare bytecode size
    forge test --gas-report              # compare per-function gas
    forge test                           # full suite — confirm no regressions

Example finding (CO-003):

CO-003 | LendingPool.sol:2 | pragma solidity ^0.8.0
  Severity : low
  Gas saved : ~30 gas/loop iteration (auto-unchecked, 0.8.22+)
              + transient storage available (0.8.24+)

  Current:
    pragma solidity ^0.8.0;

  Recommended:
    pragma solidity <latest-stable>;   // check https://github.com/ethereum/solidity/releases

  Rationale: 0.8.22+ auto-unchecks simple loop counters (no manual
  unchecked{} wrapper needed). 0.8.24+ enables native tstore/tload for
  reentrancy guards (~26,900 gas saved per guarded call vs SSTORE-based
  guard). The latest stable release includes all incremental optimizer
  improvements — look up the current version before recommending.

  Action: Bump pragma to the current stable release, run forge test for
  regressions, re-verify deployment artifacts match the new compiler version.