feat(security): harden puzzle layer params (n=128, m=192) #74
Conversation
- Increase Layer 4 puzzle parameters: - Secret dimension: n=48 → n=128 - Samples: m=72 → m=192 - Threshold: 300 → 800 - Search space: 3^48 (~2^76) → 3^128 (~2^203 brute-force) - Update contracts: TLOSWithPuzzleV5.sol, WeakLWEPuzzleV7.sol - Update examples: TLOSDeadManSwitch, TLOSRecovery, TLOSSealedAuction, TLOSTreasureHunt - Update tests with int8[128] arrays and gas threshold assertions - Update docs: AGENTS.md, README.md, docs/layers/layer4-puzzle/ - Add build artifacts to .gitignore Gas: ~8.6M puzzle verification (14% of 60M block limit) All 157 tests passing. Closes #70
|
WalkthroughThis PR increases puzzle layer security by updating LWE parameters from n=48, m=72 to n=128, m=192, and threshold from 300 to 800. Function signatures accommodating puzzle solutions expand from Changes
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~40 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 inconclusive)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. Comment |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 87082b97b9
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| function checkWithPuzzle(bytes32 input, int8[128] calldata puzzleSolution) | ||
| external |
There was a problem hiding this comment.
Because checkWithPuzzle now takes int8[128], running the documented forge script scripts/BenchmarkTLOS.s.sol (which is still hardcoded to int8[48], N_WEAK=48, M_WEAK=72) will fail to compile: the script passes a 48‑element plantedSecret into checkWithPuzzle/commit. This regression breaks the benchmark workflow unless the script’s puzzle parameters and array sizes are updated to 128/192 to match the new contract interface.
Useful? React with 👍 / 👎.
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 5
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (2)
README.md (1)
294-297: Inconsistent noise parameter in example.The lattice-estimator CLI example uses
"stddev":8.0but the documentation elsewhere states σ=25 for the LWE layer. Consider updating this example to reflect the current production parameters.Suggested fix
lattice-estimator-cli 384 65521 \ --s-dist '{"distribution":"uniform_mod"}' \ - --e-dist '{"distribution":"discrete_gaussian","stddev":8.0}' \ + --e-dist '{"distribution":"discrete_gaussian","stddev":25.0}' \ --m 2560examples/TLOSSealedAuction.sol (1)
169-175: Stale comment: puzzle solution array size should be 128, not 64.The documentation comment references
int8[64]but should beint8[128]to match the updated function signatures.📝 Suggested fix
- /// int8[64] memory puzzleSolution = solvePlantedLWE(bidSeed); // Off-chain solver + /// int8[128] memory puzzleSolution = solvePlantedLWE(bidSeed); // Off-chain solver
🤖 Fix all issues with AI agents
In @.gitignore:
- Around line 9-13: The .gitignore contains duplicate entries; remove the
repeated lines for "target/", "out/", "cache/", and "paper/*.aux" and retain
only the new "broadcast/" entry so the file no longer lists duplicates (ensure
the remaining .gitignore contains one entry per pattern and includes
"broadcast/").
In `@out/TLOSStopLoss.sol/TLOSStopLoss.json`:
- Line 1: Build artifacts under out/ (e.g.,
out/TLOSStopLoss.sol/TLOSStopLoss.json and out/Base.sol/CommonBase.json) are
still tracked despite .gitignore; remove them from git tracking and prevent
future commits by running git rm --cached <file> (or git rm -r --cached out/) to
untrack all artifacts, ensure the out/ pattern exists in .gitignore, commit the
removal with a clear message, and push the commit so subsequent builds are
ignored.
In `@out/TLOSStopLoss.t.sol/MockOracle.json`:
- Line 1: The out/ build artifact files (e.g., MockOracle.json) are still
tracked by git even though out/ is in .gitignore; stop tracking them by running
git rm --cached out/ -r to unstage and remove the tracked files from the index,
then commit the change (git commit -m "Remove generated out/ artifacts from
repo") so future builds like MockOracle.json are ignored; do not delete the
files locally so the artifacts remain for local use.
🧹 Nitpick comments (3)
test/WeakLWEPuzzleV7.t.sol (1)
8-15: Test correctly updated for expanded puzzle dimensions.The gas test now uses
int8[128]for the planted secret and the log message accurately reflects the new parameters (n=128, m=192). This provides useful gas benchmarking for the hardened puzzle.Consider adding a gas threshold assertion to catch unexpected regressions, similar to other tests mentioned in the PR (which assert gas stays within expected bounds).
💡 Optional: Add gas threshold assertion
function testGasV7() public { bytes32 x = keccak256("gas"); int8[128] memory s = puzzle.getPlantedSecret(x); uint256 g = gasleft(); (bool v,) = puzzle.verifyPuzzle(x, s); - console.log("V7 Gas (n=128, m=192):", g - gasleft()); + uint256 gasUsed = g - gasleft(); + console.log("V7 Gas (n=128, m=192):", gasUsed); + // Assert gas stays within expected bounds (~8.6M per PR documentation) + assertLt(gasUsed, 10_000_000, "Puzzle verification gas exceeded threshold"); assertTrue(v); }test/PuzzleVariants.t.sol (1)
69-76: Security level calculation is correct but could be clearer.The brute-force calculation
(128 * 1585) / 1000 ≈ 203bits is mathematically sound. Consider adding a comment that1585/1000 ≈ log₂(3)for clarity, though not required.📝 Optional: Clarify the log₂(3) approximation
function test_V7_SecurityLevel() public pure { // V7: 3^n brute-force search space = n * log2(3) ≈ n * 1.585 bits // For n=128: 128 * 1.585 ≈ 203 bits brute-force // Note: Lattice security TBD via estimator uint256 n = 128; - uint256 bruteForce_bits = (n * 1585) / 1000; // ~203 bits + uint256 bruteForce_bits = (n * 1585) / 1000; // log2(3) ≈ 1.585, so ~203 bits assert(bruteForce_bits >= 128); }docs/layers/layer4-puzzle/README.md (1)
44-66: Code snippet is illustrative but differs from actual implementation.The code snippet shows pseudocode (
residual²,mod q) rather than actual Solidity. This is acceptable for documentation purposes, but consider adding a note that this is simplified pseudocode, or updating to match the actual implementation more closely.
📜 Review details
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (41)
.gitignoreAGENTS.mdREADME.mdcache/solidity-files-cache.jsoncontracts/TLOSWithPuzzleV5.solcontracts/WeakLWEPuzzleV7.soldocs/layers/layer4-puzzle/README.mdexamples/TLOSDeadManSwitch.solexamples/TLOSRecovery.solexamples/TLOSSealedAuction.solexamples/TLOSStopLoss.solexamples/TLOSTreasureHunt.solout/IHoneypot.sol/IHoneypot.jsonout/TLOSDeadManSwitch.sol/TLOSDeadManSwitch.jsonout/TLOSDeadManSwitch.t.sol/TLOSDeadManSwitchTest.jsonout/TLOSRecovery.sol/TLOSRecovery.jsonout/TLOSRecovery.t.sol/TLOSRecoveryTest.jsonout/TLOSSealedAuction.sol/TLOSSealedAuction.jsonout/TLOSSealedAuction.t.sol/TLOSSealedAuctionTest.jsonout/TLOSStopLoss.sol/IOracle.jsonout/TLOSStopLoss.sol/ITLOSCircuit.jsonout/TLOSStopLoss.sol/TLOSStopLoss.jsonout/TLOSStopLoss.t.sol/MockOracle.jsonout/TLOSStopLoss.t.sol/MockTLOSCircuit.jsonout/TLOSStopLoss.t.sol/TLOSStopLossTest.jsonout/TLOSTreasureHunt.sol/TLOSTreasureHunt.jsonout/TLOSTreasureHunt.t.sol/TLOSTreasureHuntTest.jsonout/WeakLWEPuzzleV7.sol/WeakLWEPuzzleV7.jsonout/WeakLWEPuzzleV7.t.sol/V7GasTest.jsonout/build-info/5f146a5d33a41b8e.jsonout/build-info/9163372221981f5a.jsonout/build-info/b32938c0fe325972.jsonout/build-info/d5624d0f4ad6ffe6.jsontest/PuzzleVariants.t.soltest/TLOSDeadManSwitch.t.soltest/TLOSRecovery.t.soltest/TLOSSealedAuction.t.soltest/TLOSTreasureHunt.t.soltest/TLOSWithPuzzleV5.t.soltest/TLOSWithPuzzleV5Harness.soltest/WeakLWEPuzzleV7.t.sol
💤 Files with no reviewable changes (4)
- out/build-info/9163372221981f5a.json
- out/build-info/5f146a5d33a41b8e.json
- out/build-info/d5624d0f4ad6ffe6.json
- out/build-info/b32938c0fe325972.json
🧰 Additional context used
📓 Path-based instructions (9)
**/*.sol
📄 CodeRabbit inference engine (AGENTS.md)
**/*.sol: Use Forge for building and testing Solidity contracts with commands:forge buildfor compilation andforge testfor running tests
Solidity contracts must use seed-derived a vectors with 11 bytes per gate for storage optimization
Files:
test/WeakLWEPuzzleV7.t.soltest/TLOSWithPuzzleV5Harness.soltest/TLOSSealedAuction.t.soltest/PuzzleVariants.t.solcontracts/WeakLWEPuzzleV7.soltest/TLOSDeadManSwitch.t.solexamples/TLOSTreasureHunt.soltest/TLOSTreasureHunt.t.solexamples/TLOSRecovery.solexamples/TLOSSealedAuction.soltest/TLOSRecovery.t.solexamples/TLOSDeadManSwitch.solcontracts/TLOSWithPuzzleV5.solexamples/TLOSStopLoss.soltest/TLOSWithPuzzleV5.t.sol
{**/*.sol,src/lwe.rs}
📄 CodeRabbit inference engine (AGENTS.md)
Production LWE layer must use parameters: n=384 dimension, σ=8 Gaussian noise, q=65521 modulus, achieving ~2^112 PQ security
Files:
test/WeakLWEPuzzleV7.t.soltest/TLOSWithPuzzleV5Harness.soltest/TLOSSealedAuction.t.soltest/PuzzleVariants.t.solcontracts/WeakLWEPuzzleV7.soltest/TLOSDeadManSwitch.t.solexamples/TLOSTreasureHunt.soltest/TLOSTreasureHunt.t.solexamples/TLOSRecovery.solexamples/TLOSSealedAuction.soltest/TLOSRecovery.t.solexamples/TLOSDeadManSwitch.solcontracts/TLOSWithPuzzleV5.solexamples/TLOSStopLoss.soltest/TLOSWithPuzzleV5.t.sol
**/{WeakLWEPuzzleV7,*Puzzle*}.sol
📄 CodeRabbit inference engine (AGENTS.md)
Production Layer 4 puzzle (WeakLWEPuzzleV7) must use: n=48 secret dimension, m=72 samples, q=2039 modulus, error range {-2,-1,0,1,2}, threshold=300, ensuring 3^48 ≈ 2^76 search space
Files:
test/WeakLWEPuzzleV7.t.soltest/TLOSWithPuzzleV5Harness.soltest/PuzzleVariants.t.solcontracts/WeakLWEPuzzleV7.solcontracts/TLOSWithPuzzleV5.soltest/TLOSWithPuzzleV5.t.sol
{**/*.sol,src/wire_binding.rs}
📄 CodeRabbit inference engine (AGENTS.md)
Layer 3 wire binding must implement full-rank 64x64 linear hash over Z_q for inter-gate algebraic consistency; do NOT claim collision resistance as the linear system is trivially solvable
Files:
test/WeakLWEPuzzleV7.t.soltest/TLOSWithPuzzleV5Harness.soltest/TLOSSealedAuction.t.soltest/PuzzleVariants.t.solcontracts/WeakLWEPuzzleV7.soltest/TLOSDeadManSwitch.t.solexamples/TLOSTreasureHunt.soltest/TLOSTreasureHunt.t.solexamples/TLOSRecovery.solexamples/TLOSSealedAuction.soltest/TLOSRecovery.t.solexamples/TLOSDeadManSwitch.solcontracts/TLOSWithPuzzleV5.solexamples/TLOSStopLoss.soltest/TLOSWithPuzzleV5.t.sol
test/*.t.sol
📄 CodeRabbit inference engine (AGENTS.md)
Test contracts must include comprehensive layer-specific tests: production contract tests in TLOSWithPuzzleV5.t.sol (61 tests), isolated layer testing via TLOSWithPuzzleV5Harness.sol, and puzzle variant tests in PuzzleVariants.t.sol (12 tests covering V5/V6/V7)
Files:
test/WeakLWEPuzzleV7.t.soltest/TLOSSealedAuction.t.soltest/PuzzleVariants.t.soltest/TLOSDeadManSwitch.t.soltest/TLOSTreasureHunt.t.soltest/TLOSRecovery.t.soltest/TLOSWithPuzzleV5.t.sol
contracts/WeakLWEPuzzle*.sol
📄 CodeRabbit inference engine (AGENTS.md)
Testing puzzle variants must include WeakLWEPuzzleV5 (n=32, 2^51 security), WeakLWEPuzzleV6 (n=24, 2^38 security), and WeakLWEPuzzleV7 (n=48, 2^76 security) for parameter validation
Files:
contracts/WeakLWEPuzzleV7.sol
contracts/*.sol
📄 CodeRabbit inference engine (AGENTS.md)
contracts/*.sol: The Ethereum block gas limit is 60,000,000 gas (60M) as of 2024; TLOS gas usage ranges from 3.7M-17.2M (6-28% of block limit) with batch size of 128 gates and 5 binding updates for 640 gates
Hash-PoW layer must provide commit-time randomness bound with configurable difficulty and default enabled state for Layer 5 security
Files:
contracts/WeakLWEPuzzleV7.solcontracts/TLOSWithPuzzleV5.sol
examples/**/*.sol
📄 CodeRabbit inference engine (AGENTS.md)
Demo/example contracts in examples/ directory are for educational purposes only and must not be used in production
Files:
examples/TLOSTreasureHunt.solexamples/TLOSRecovery.solexamples/TLOSSealedAuction.solexamples/TLOSDeadManSwitch.solexamples/TLOSStopLoss.sol
docs/layers/**
📄 CodeRabbit inference engine (AGENTS.md)
Documentation must be organized per-layer in docs/layers/ directory for technical reference of each security layer
Files:
docs/layers/layer4-puzzle/README.md
🧠 Learnings (17)
📓 Common learnings
Learnt from: igor53627
Repo: igor53627/tlos PR: 0
File: :0-0
Timestamp: 2026-01-15T08:50:51.160Z
Learning: Production Layer 4 puzzle (WeakLWEPuzzleV7) must use: n=64 secret dimension, m=96 samples, q=2039 modulus, error range {-2,-1,0,1,2}, threshold=400, ensuring 3^64 ≈ 2^101 search space, 2.30M gas (Tenderly, production-style).
Learnt from: igor53627
Repo: igor53627/tlos PR: 0
File: :0-0
Timestamp: 2026-01-15T08:50:51.160Z
Learning: Applies to contracts/*.sol : TLOS gas usage with n=384 ranges from 4,734,943-18,184,574 gas (7-30% of 60M block limit) for 64-640 gates based on Tenderly benchmarks; includes 2.30M puzzle verification.
Learnt from: igor53627
Repo: igor53627/tlos PR: 0
File: :0-0
Timestamp: 2026-01-15T08:50:51.160Z
Learning: Five-layer security model must be implemented in order: (1) Topology layer - structural mixing, (2) LWE layer - control function hiding, (3) Wire binding layer - algebraic binding, (4) Planted LWE puzzle - 2^101 search space, (5) Hash-PoW - commit-time randomness (default enabled).
Learnt from: CR
Repo: igor53627/tlos PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-01-13T11:05:17.527Z
Learning: Applies to test/*.t.sol : Test contracts must include comprehensive layer-specific tests: production contract tests in TLOSWithPuzzleV5.t.sol (61 tests), isolated layer testing via TLOSWithPuzzleV5Harness.sol, and puzzle variant tests in PuzzleVariants.t.sol (12 tests covering V5/V6/V7)
Learnt from: igor53627
Repo: igor53627/tlos PR: 0
File: :0-0
Timestamp: 2026-01-15T08:50:51.160Z
Learning: Production LWE layer must use parameters: n=384 dimension, σ=25 Gaussian noise, q=65521 modulus, achieving ~2^112 PQ security.
📚 Learning: 2026-01-15T08:50:51.160Z
Learnt from: igor53627
Repo: igor53627/tlos PR: 0
File: :0-0
Timestamp: 2026-01-15T08:50:51.160Z
Learning: Production Layer 4 puzzle (WeakLWEPuzzleV7) must use: n=64 secret dimension, m=96 samples, q=2039 modulus, error range {-2,-1,0,1,2}, threshold=400, ensuring 3^64 ≈ 2^101 search space, 2.30M gas (Tenderly, production-style).
Applied to files:
test/WeakLWEPuzzleV7.t.soltest/TLOSWithPuzzleV5Harness.soltest/TLOSSealedAuction.t.solREADME.mdtest/PuzzleVariants.t.solout/WeakLWEPuzzleV7.sol/WeakLWEPuzzleV7.jsoncontracts/WeakLWEPuzzleV7.soltest/TLOSDeadManSwitch.t.solexamples/TLOSTreasureHunt.solAGENTS.mdtest/TLOSTreasureHunt.t.solexamples/TLOSRecovery.solexamples/TLOSSealedAuction.soltest/TLOSRecovery.t.solexamples/TLOSDeadManSwitch.solout/TLOSTreasureHunt.sol/TLOSTreasureHunt.jsoncontracts/TLOSWithPuzzleV5.solexamples/TLOSStopLoss.soldocs/layers/layer4-puzzle/README.mdtest/TLOSWithPuzzleV5.t.sol
📚 Learning: 2026-01-13T11:05:17.527Z
Learnt from: CR
Repo: igor53627/tlos PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-01-13T11:05:17.527Z
Learning: Applies to test/*.t.sol : Test contracts must include comprehensive layer-specific tests: production contract tests in TLOSWithPuzzleV5.t.sol (61 tests), isolated layer testing via TLOSWithPuzzleV5Harness.sol, and puzzle variant tests in PuzzleVariants.t.sol (12 tests covering V5/V6/V7)
Applied to files:
test/WeakLWEPuzzleV7.t.soltest/TLOSWithPuzzleV5Harness.soltest/TLOSSealedAuction.t.solout/TLOSStopLoss.t.sol/MockOracle.jsonREADME.mdtest/PuzzleVariants.t.solout/WeakLWEPuzzleV7.sol/WeakLWEPuzzleV7.jsoncontracts/WeakLWEPuzzleV7.soltest/TLOSDeadManSwitch.t.solexamples/TLOSTreasureHunt.solout/TLOSRecovery.sol/TLOSRecovery.jsonAGENTS.mdtest/TLOSTreasureHunt.t.solexamples/TLOSRecovery.solexamples/TLOSSealedAuction.soltest/TLOSRecovery.t.solexamples/TLOSDeadManSwitch.solout/TLOSTreasureHunt.sol/TLOSTreasureHunt.jsoncontracts/TLOSWithPuzzleV5.solcache/solidity-files-cache.jsondocs/layers/layer4-puzzle/README.mdtest/TLOSWithPuzzleV5.t.sol
📚 Learning: 2026-01-15T08:50:51.160Z
Learnt from: igor53627
Repo: igor53627/tlos PR: 0
File: :0-0
Timestamp: 2026-01-15T08:50:51.160Z
Learning: Applies to contracts/*.sol : TLOS gas usage with n=384 ranges from 4,734,943-18,184,574 gas (7-30% of 60M block limit) for 64-640 gates based on Tenderly benchmarks; includes 2.30M puzzle verification.
Applied to files:
test/WeakLWEPuzzleV7.t.soltest/TLOSSealedAuction.t.solREADME.mdtest/PuzzleVariants.t.solcontracts/WeakLWEPuzzleV7.soltest/TLOSDeadManSwitch.t.solexamples/TLOSTreasureHunt.solAGENTS.mdtest/TLOSTreasureHunt.t.solexamples/TLOSRecovery.solexamples/TLOSSealedAuction.soltest/TLOSRecovery.t.solout/TLOSStopLoss.sol/IOracle.jsonexamples/TLOSDeadManSwitch.solcontracts/TLOSWithPuzzleV5.solcache/solidity-files-cache.jsondocs/layers/layer4-puzzle/README.mdtest/TLOSWithPuzzleV5.t.sol
📚 Learning: 2026-01-15T08:50:51.160Z
Learnt from: igor53627
Repo: igor53627/tlos PR: 0
File: :0-0
Timestamp: 2026-01-15T08:50:51.160Z
Learning: Five-layer security model must be implemented in order: (1) Topology layer - structural mixing, (2) LWE layer - control function hiding, (3) Wire binding layer - algebraic binding, (4) Planted LWE puzzle - 2^101 search space, (5) Hash-PoW - commit-time randomness (default enabled).
Applied to files:
README.mdout/WeakLWEPuzzleV7.sol/WeakLWEPuzzleV7.jsoncontracts/WeakLWEPuzzleV7.solexamples/TLOSTreasureHunt.solAGENTS.mdexamples/TLOSSealedAuction.solexamples/TLOSDeadManSwitch.solcontracts/TLOSWithPuzzleV5.solexamples/TLOSStopLoss.soldocs/layers/layer4-puzzle/README.mdtest/TLOSWithPuzzleV5.t.sol
📚 Learning: 2026-01-15T08:50:51.160Z
Learnt from: igor53627
Repo: igor53627/tlos PR: 0
File: :0-0
Timestamp: 2026-01-15T08:50:51.160Z
Learning: Production LWE layer must use parameters: n=384 dimension, σ=25 Gaussian noise, q=65521 modulus, achieving ~2^112 PQ security.
Applied to files:
README.mdAGENTS.mdcontracts/TLOSWithPuzzleV5.soldocs/layers/layer4-puzzle/README.md
📚 Learning: 2026-01-13T11:05:17.527Z
Learnt from: CR
Repo: igor53627/tlos PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-01-13T11:05:17.527Z
Learning: Applies to {**/*.sol,src/wire_binding.rs} : Layer 3 wire binding must implement full-rank 64x64 linear hash over Z_q for inter-gate algebraic consistency; do NOT claim collision resistance as the linear system is trivially solvable
Applied to files:
README.mdAGENTS.mdcontracts/TLOSWithPuzzleV5.solexamples/TLOSStopLoss.sol
📚 Learning: 2026-01-13T11:05:17.527Z
Learnt from: CR
Repo: igor53627/tlos PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-01-13T11:05:17.527Z
Learning: Applies to contracts/*.sol : The Ethereum block gas limit is 60,000,000 gas (60M) as of 2024; TLOS gas usage ranges from 3.7M-17.2M (6-28% of block limit) with batch size of 128 gates and 5 binding updates for 640 gates
Applied to files:
README.mdAGENTS.mdout/TLOSStopLoss.sol/IOracle.jsondocs/layers/layer4-puzzle/README.md
📚 Learning: 2026-01-13T11:05:17.527Z
Learnt from: CR
Repo: igor53627/tlos PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-01-13T11:05:17.527Z
Learning: Applies to contracts/*.sol : Hash-PoW layer must provide commit-time randomness bound with configurable difficulty and default enabled state for Layer 5 security
Applied to files:
README.mdcontracts/WeakLWEPuzzleV7.solexamples/TLOSTreasureHunt.solAGENTS.mdexamples/TLOSSealedAuction.solexamples/TLOSDeadManSwitch.solcontracts/TLOSWithPuzzleV5.solout/IHoneypot.sol/IHoneypot.json
📚 Learning: 2026-01-13T11:05:17.527Z
Learnt from: CR
Repo: igor53627/tlos PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-01-13T11:05:17.527Z
Learning: Applies to src/*.rs : Rust implementation must have modular structure: circuit.rs for topology layer mixing, lwe.rs for LWE encryption, wire_binding.rs for algebraic binding, and generator.rs for deployment data generation
Applied to files:
README.mdAGENTS.md
📚 Learning: 2026-01-13T11:05:17.527Z
Learnt from: CR
Repo: igor53627/tlos PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-01-13T11:05:17.527Z
Learning: Applies to docs/layers/** : Documentation must be organized per-layer in docs/layers/ directory for technical reference of each security layer
Applied to files:
README.md
📚 Learning: 2026-01-13T11:05:17.527Z
Learnt from: CR
Repo: igor53627/tlos PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-01-13T11:05:17.527Z
Learning: Applies to scripts/Benchmark*.s.sol : Benchmark scripts must use Tenderly RPC by sourcing `~/.zsh_secrets` and running: `forge script scripts/BenchmarkTLOS.s.sol --rpc-url "$TENDERLY_RPC" --broadcast --unlocked -vvv`
Applied to files:
README.md
📚 Learning: 2026-01-13T11:05:17.527Z
Learnt from: CR
Repo: igor53627/tlos PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-01-13T11:05:17.527Z
Learning: Applies to examples/**/*.sol : Demo/example contracts in examples/ directory are for educational purposes only and must not be used in production
Applied to files:
cache/solidity-files-cache.json
📚 Learning: 2026-01-13T11:05:17.527Z
Learnt from: CR
Repo: igor53627/tlos PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-01-13T11:05:17.527Z
Learning: Applies to **/*.sol : Use Forge for building and testing Solidity contracts with commands: `forge build` for compilation and `forge test` for running tests
Applied to files:
cache/solidity-files-cache.json
📚 Learning: 2026-01-13T11:05:17.527Z
Learnt from: CR
Repo: igor53627/tlos PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-01-13T11:05:17.527Z
Learning: Applies to paper/*.tex : Build papers by running pdflatex twice for references and bibtex for citations; check for undefined citations with: `pdflatex <file>.tex 2>&1 | grep -i "undefined|Citation"`; if citations show as "[?]" verify refs.bib contains entry and re-run bibtex then pdflatex twice
Applied to files:
.gitignore
📚 Learning: 2026-01-13T11:05:17.527Z
Learnt from: CR
Repo: igor53627/tlos PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-01-13T11:05:17.527Z
Learning: Applies to paper/** : DO NOT create markdown (.md) files for papers; use LaTeX only (tlos.tex for short papers, tlos-paper.tex for full papers)
Applied to files:
.gitignore
📚 Learning: 2026-01-13T11:05:17.527Z
Learnt from: CR
Repo: igor53627/tlos PR: 0
File: AGENTS.md:0-0
Timestamp: 2026-01-13T11:05:17.527Z
Learning: Applies to **/*.sol : Solidity contracts must use seed-derived a vectors with 11 bytes per gate for storage optimization
Applied to files:
test/TLOSWithPuzzleV5.t.sol
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: Cursor Bugbot
🔇 Additional comments (58)
out/IHoneypot.sol/IHoneypot.json (1)
1-1: No action needed—out/is already properly configured in.gitignoreand this file is not being modified in the current changeset.The file appears in the review as part of existing repository history, not as a newly staged artifact. Since build artifacts are correctly excluded from version control via
.gitignore, there is no risk of accidental re-commitment.examples/TLOSStopLoss.sol (1)
81-81: Documentation update aligns with PR objectives.The comment correctly reflects the increased puzzle parameters (3^128 brute-force search space) and honestly notes that lattice security analysis is pending. This is appropriate for a demo contract.
out/TLOSStopLoss.sol/IOracle.json (1)
1-1: Auto-generated artifact — no functional concerns.This is a compilation artifact for the
IOracleinterface. The embedded documentation correctly reflects the updated puzzle parameters. Same gitignore concern applies as noted for otherout/artifacts.out/WeakLWEPuzzleV7.sol/WeakLWEPuzzleV7.json (1)
1-1: ABI correctly reflects expanded puzzle dimensions (n=128).The artifact shows
int8[128]arrays forgetPlantedSecretandverifyPuzzle, matching the PR's parameter changes. The embedded devdoc accurately documents "n=128, m=192, q=2039, threshold=800".Note: The original issue
#70proposed n=64, but this PR implements n=128. Ensure this more aggressive hardening was intentionally chosen over the original proposal, considering the increased gas cost (~8.6M vs ~2.3M estimated for n=64).AGENTS.md (2)
75-81: Layer 4 Puzzle Parameters table looks correct for the PR changes.The table accurately reflects the hardened puzzle parameters:
- n=128, m=192, q=2039, threshold=800
- 3^128 brute-force search space with lattice security TBD
- 8.62M verification gas (14% of 60M block limit)
This table is consistent with the PR objectives and the build artifact ABIs reviewed earlier.
9-12: Internal documentation inconsistencies need resolution.The five-layer security model description and repository structure sections are inconsistent with the Layer 4 Puzzle Parameters table:
Section Parameter Conflict Line 11 "3^64 ≈ 2^101" and "2.30M gas" Line 80-81 says "3^128" and "8.62M gas" Line 35 "n=64, 2^101 security" Line 75 says "n = 128" Line 34 "puzzle n=64" Line 75-80 shows n=128, m=192 These should all reflect the new parameters (n=128, m=192, threshold=800, 3^128 search space, ~8.6M gas).
📝 Suggested fixes for consistency
Five-layer security model: 1. **Topology layer** - structural mixing (heuristic) 2. **LWE layer** - control function hiding via standard LWE with Gaussian noise (σ=25, n=384, ~2^112 PQ) 3. **Wire binding layer** - full-rank linear hash for inter-gate consistency (algebraic binding, inspired by [MDS25]) -4. **Planted LWE puzzle** - forces minimum 3^64 ≈ 2^101 brute-force search space (2.30M gas (Tenderly)) +4. **Planted LWE puzzle** - forces minimum 3^128 ≈ 2^203 brute-force search space (8.62M gas (Tenderly)) 5. **Hash-PoW** - commit-time randomness bound (configurable difficulty; default enabled)- `contracts/` - Solidity contracts - - `TLOSWithPuzzleV5.sol` - Production: 5-layer TLOS (n=384, puzzle n=64) - - `WeakLWEPuzzleV7.sol` - Gas Testing: Hardened puzzle (n=64, 2^101 security, deterministic noise) + - `TLOSWithPuzzleV5.sol` - Production: 5-layer TLOS (n=384, puzzle n=128) + - `WeakLWEPuzzleV7.sol` - Gas Testing: Hardened puzzle (n=128, 2^203 brute-force, deterministic noise)⛔ Skipped due to learnings
Learnt from: igor53627 Repo: igor53627/tlos PR: 0 File: :0-0 Timestamp: 2026-01-15T08:50:51.160Z Learning: Production Layer 4 puzzle (WeakLWEPuzzleV7) must use: n=64 secret dimension, m=96 samples, q=2039 modulus, error range {-2,-1,0,1,2}, threshold=400, ensuring 3^64 ≈ 2^101 search space, 2.30M gas (Tenderly, production-style).Learnt from: igor53627 Repo: igor53627/tlos PR: 0 File: :0-0 Timestamp: 2026-01-15T08:50:51.160Z Learning: Five-layer security model must be implemented in order: (1) Topology layer - structural mixing, (2) LWE layer - control function hiding, (3) Wire binding layer - algebraic binding, (4) Planted LWE puzzle - 2^101 search space, (5) Hash-PoW - commit-time randomness (default enabled).Learnt from: CR Repo: igor53627/tlos PR: 0 File: AGENTS.md:0-0 Timestamp: 2026-01-13T11:05:17.527Z Learning: Applies to test/*.t.sol : Test contracts must include comprehensive layer-specific tests: production contract tests in TLOSWithPuzzleV5.t.sol (61 tests), isolated layer testing via TLOSWithPuzzleV5Harness.sol, and puzzle variant tests in PuzzleVariants.t.sol (12 tests covering V5/V6/V7)Learnt from: igor53627 Repo: igor53627/tlos PR: 0 File: :0-0 Timestamp: 2026-01-15T08:50:51.160Z Learning: Production LWE layer must use parameters: n=384 dimension, σ=25 Gaussian noise, q=65521 modulus, achieving ~2^112 PQ security.Learnt from: igor53627 Repo: igor53627/tlos PR: 0 File: :0-0 Timestamp: 2026-01-15T08:50:51.160Z Learning: Applies to contracts/*.sol : TLOS gas usage with n=384 ranges from 4,734,943-18,184,574 gas (7-30% of 60M block limit) for 64-640 gates based on Tenderly benchmarks; includes 2.30M puzzle verification.Learnt from: CR Repo: igor53627/tlos PR: 0 File: AGENTS.md:0-0 Timestamp: 2026-01-13T11:05:17.527Z Learning: Applies to contracts/*.sol : Hash-PoW layer must provide commit-time randomness bound with configurable difficulty and default enabled state for Layer 5 securityLearnt from: CR Repo: igor53627/tlos PR: 0 File: AGENTS.md:0-0 Timestamp: 2026-01-13T11:05:17.527Z Learning: Applies to docs/layers/** : Documentation must be organized per-layer in docs/layers/ directory for technical reference of each security layerLearnt from: CR Repo: igor53627/tlos PR: 0 File: AGENTS.md:0-0 Timestamp: 2026-01-13T11:05:17.527Z Learning: Applies to {**/*.sol,src/wire_binding.rs} : Layer 3 wire binding must implement full-rank 64x64 linear hash over Z_q for inter-gate algebraic consistency; do NOT claim collision resistance as the linear system is trivially solvablecontracts/WeakLWEPuzzleV7.sol (3)
4-15: Parameter hardening looks correct, but coding guidelines need updating.The new parameters (n=128, m=192, threshold=800) align with the PR objectives for Issue
#70. The threshold comment correctly notesE[||e||²] ≈ 192 * 2 = 384, and the 800 threshold provides ~2x headroom for valid solutions.However, the repository's coding guidelines still specify:
Production Layer 4 puzzle must use: n=48, m=72, threshold=300
Consider updating the coding guidelines to reflect these hardened parameters after this PR merges.
36-48: Dynamic block count calculation is correct.The formula
(N_WEAK + 15) / 16correctly computes 8 blocks for n=128, ensuring all 128 secret elements are processed. The inner loop boundary checkif (idx >= N_WEAK) breakproperly handles the exact 128 elements.
97-112: LGTM!The
getPlantedSecretfunction correctly mirrors the planted secret generation inverifyPuzzle, with the return type properly updated toint8[128].README.md (2)
132-145: Documentation accurately reflects implementation.The parameter table correctly documents n=128, m=192, threshold=800, and the 8.62M gas cost. The note about "lattice security TBD via estimator" appropriately acknowledges that while brute-force is 3^128, formal lattice analysis is pending.
61-78: Gas documentation is consistent with implementation.The gas table correctly reflects the increased puzzle verification cost (~8.6M) included in the total gas figures.
out/TLOSRecovery.sol/TLOSRecovery.json (1)
1-1: Build artifact correctly reflects updated signatures.The ABI shows the expanded
int8[128]array sizes forcheckRecovery,getPlantedSecret, andrecoverfunctions, consistent with the WeakLWEPuzzleV7 parameter changes.cache/solidity-files-cache.json (1)
1-1: Build cache reflects successful compilation.This auto-generated Forge cache file correctly tracks the updated source files and their compilation artifacts. No issues identified.
out/TLOSStopLoss.t.sol/MockTLOSCircuit.json (1)
1-1: Test mock artifact unchanged functionally.The MockTLOSCircuit ABI remains unchanged (check, setResult, shouldReturn). The file was recompiled alongside other changes, updating only metadata/sourceMap. No issues.
contracts/TLOSWithPuzzleV5.sol (2)
235-254: Dynamic block calculation correctly handles expanded dimensions.The block iteration logic
(N_WEAK + 15) / 16correctly computes the ceiling division, yielding 8 blocks for N_WEAK=128. This generalizes the previous fixed 3-block loop and will correctly handle any future parameter changes.
128-137: Commitment hash includes full 128-element puzzle solution.The commitment in
revealWithPuzzlecorrectly encodes the larger puzzle solution array, ensuring the binding between commit and reveal remains intact for the expanded parameter space.examples/TLOSRecovery.sol (2)
22-25: Example contract parameters aligned with production.The puzzle parameters in this educational example match the production contract (TLOSWithPuzzleV5.sol), ensuring consistency for developers learning from the examples. As per coding guidelines, this contract is for demonstration purposes only.
119-134: Planted secret generation uses correct 128-element array.The
int16[128] memory plantedarray and dynamic block iteration correctly handle the expanded puzzle dimensions. The on-the-fly derivation approach is appropriate for this educational example.examples/TLOSSealedAuction.sol (2)
61-64: Puzzle parameters correctly updated for auction example.Constants align with the production contract and other examples, maintaining consistency across the codebase.
408-437: Puzzle verification correctly implements expanded 128-dimension parameters.The
_verifyPuzzlefunction correctly handles the larger planted secret and solution arrays with dynamic block iteration.out/TLOSTreasureHunt.sol/TLOSTreasureHunt.json (1)
1-1: Build artifact correctly reflects source code changes.The ABI updates (
getPlantedSecretreturningint8[128],revealacceptingint8[128]) are consistent with the puzzle parameter expansion in the source contract. Since this is an example contract underexamples/, the changes are appropriate for demonstrating the hardened puzzle parameters.test/TLOSTreasureHunt.t.sol (3)
37-47: LGTM!The puzzle solution array size correctly updated to
int8[128]to match the contract's updatedgetPlantedSecretreturn type andrevealparameter.
93-108: LGTM!The wrong solution test correctly updates both the array declaration and loop bound to 128 elements, ensuring full coverage of the expanded puzzle dimensions.
208-224: LGTM!The gas measurement test with a 15M threshold provides adequate headroom above the reported ~8.6M actual usage while catching unexpected gas regressions. The assertion message correctly documents the
n=128puzzle context.test/TLOSDeadManSwitch.t.sol (4)
99-108: LGTM!The puzzle solution array correctly updated to
int8[128]for the heir claim test.
155-173: LGTM!The wrong puzzle test correctly initializes a 128-element array with non-zero values to test invalid solution rejection.
175-216: LGTM!Multiple heir claim test correctly updates both solution arrays to
int8[128], maintaining comprehensive coverage of the commit-reveal flow with expanded puzzle dimensions.
246-266: LGTM!Gas measurement test correctly updated with the 128-element solution array and a consistent 15M gas threshold assertion matching other puzzle verification tests in the suite.
out/TLOSStopLoss.sol/ITLOSCircuit.json (1)
1-1: LGTM!Build artifact metadata updated to reflect source compilation changes. The interface ABI remains unchanged (
check(bytes32) -> bool).test/TLOSWithPuzzleV5Harness.sol (1)
34-37: Test harness signature is correct and aligns with production implementation.The test harness
verifyPuzzlePublicusesint8[128]which correctly matches the actual WeakLWEPuzzleV7 production parameters (n=128, m=192, threshold=800) documented in AGENTS.md. The coding guidelines provided contain outdated puzzle parameters (n=48, m=72, threshold=300) that do not reflect the current implementation. No changes are required.Likely an incorrect or invalid review comment.
test/TLOSSealedAuction.t.sol (1)
46-47: LGTM - Puzzle array sizes consistently updated.All
int8[48]arrays correctly updated toint8[128]across test cases, and the loop bound intestRevealBidWrongPuzzle(line 105) correctly iterates 128 times to match the new array size. The helper function_getPuzzleSolutionreturn type is also properly updated.Also applies to: 73-74, 95-96, 104-107, 117-118, 136-137, 165-166, 193-194, 233-234, 273-274, 301-302, 320-323
test/TLOSRecovery.t.sol (2)
28-28: LGTM - Puzzle arrays and iteration bounds correctly updated.All puzzle solution arrays updated to
int8[128], and loops correctly iterate 128 times. The deterministic test at lines 154-157 properly validates all 128 elements of the ternary secret.Also applies to: 38-38, 48-51, 62-62, 69-70, 92-92, 103-103, 116-116, 133-133, 151-157
141-141: Gas threshold appropriately increased for larger puzzle.The threshold increase from 2M to 10M gas aligns with the PR's documented ~8.6M gas for
n=128puzzle verification (~14% of 60M block limit).test/PuzzleVariants.t.sol (3)
8-8: LGTM - Parameters correctly updated.The contract documentation and constant assertions properly reflect the new production parameters:
n=128,m=192,q=2039,threshold=800.Also applies to: 18-25
27-36: LGTM - Test cases properly updated for 128-element arrays.Planted secret validation, random secret rejection, all-zeros check, and non-ternary rejection tests all correctly use
int8[128]arrays with appropriate helper functions.Also applies to: 39-56
80-90: LGTM - Helper functions correctly updated.Both
_assertTernary128and_randomTernary128properly handle 128-element arrays with correct iteration bounds.docs/layers/layer4-puzzle/README.md (3)
3-3: LGTM - Documentation accurately reflects new parameters.The introduction and parameter table correctly document
n=128,m=192,threshold=800, and the3^128brute-force search space. The hedging on lattice security ("TBD via estimator") is appropriately cautious.Also applies to: 9-14
75-77: LGTM - Variant table correctly updated.V7 row properly shows
n=128,m=192,3^128 brute-force,8.62M gas, andProductionstatus. The note about lattice security requiring estimator analysis is appropriately included.
81-84: LGTM - Attack economics appropriately updated.The description correctly notes that
3^128 (~2^203 bits)brute-force is computationally infeasible and appropriately defers lattice attack claims to estimator analysis.examples/TLOSDeadManSwitch.sol (6)
26-29: LGTM - Security model comment correctly updated.The comment accurately describes the
3^128brute-force search space and appropriately notes that lattice security is TBD and low-entropy codes remain dictionary-bound.
91-95: LGTM - Constants correctly updated.
N_WEAK=128,M_WEAK=192, andPUZZLE_THRESHOLD_SQ=800match the production parameters defined in this PR.
291-296: LGTM -claim()signature correctly updated.The
puzzleSolutionparameter correctly usesint8[128] calldata.
391-406: LGTM -getPlantedSecret()correctly generates 128-element secret.The dynamic block calculation
(N_WEAK + 15) / 16 = 8correctly iterates to generate all 128 ternary secret elements. The inner loop properly breaks whenidx >= N_WEAK.
421-422: LGTM -_verifyPuzzle()signature and planted array correctly updated.The function signature uses
int8[128] calldataand the internalplantedarray is correctly sized toint16[128]. Block iteration generates all 128 planted secret elements.Also applies to: 436-451
455-490: LGTM - Verification loop correctly iterates over all samples.The outer loop iterates
M_WEAK=192times for each sample row. The inner block loop correctly uses the dynamicblocksvalue to compute dot products for all 128 columns. Residual calculation and threshold check are unchanged and correct.test/TLOSWithPuzzleV5.t.sol (8)
85-91: LGTM!The
_commitHashhelper correctly updated to acceptint8[128] memory solution.
111-122: LGTM!The
_powNonceForhelper correctly updated to acceptint8[128] memory solution.
220-238: LGTM!The single-bit-flip test correctly uses
int8[128] memory flippedSecretand validates that even a single coordinate change exceeds the threshold.
777-788: LGTM!The cross-layer test correctly iterates over
N_WEAKelements to create a negated puzzle solution.
1027-1039: LGTM!The
_generatePlantedSecrethelper correctly:
- Returns
int8[128] memory secret- Computes block count dynamically:
(N_WEAK + 15) / 16= 8 blocks for N_WEAK=128- Includes boundary check
if (idx >= N_WEAK) breakto handle non-multiple-of-16 dimensions
1041-1076: LGTM!The
_generateBVectorhelper correctly:
- Accepts
int8[128] memory secret- Allocates
M_WEAK * 2= 384 bytes for 192 rows of uint16 values- Uses dynamic block calculation for the dot product computation
- Includes proper boundary checks
956-967: LGTM!The gas threshold of 12M provides reasonable headroom above the expected ~8.6M for puzzle verification with the new n=128, m=192 parameters.
21-25: Parameter mismatch with coding guidelines and learnings—requires clarification on intended security posture.The implementation uses n=128, m=192, threshold=800, but this contradicts:
- Coding guidelines specifying n=48, m=72, threshold=300
- Learnings specifying n=64, m=96, threshold=400
The WeakLWEPuzzleV7 contract explicitly labels this n=128 variant as "Hardened" with 3^128 search space. Clarify whether this parameter escalation is an intentional security enhancement or an oversight, then update the coding guidelines to reflect the authoritative specification.
examples/TLOSTreasureHunt.sol (5)
54-57: LGTM!Constants correctly updated to match the new puzzle parameters:
N_WEAK = 128(secret dimension)M_WEAK = 192(samples)PUZZLE_THRESHOLD_SQ = 800(calibrated threshold)
140-167: LGTM!The
revealfunction correctly updated withint8[128] calldata puzzleSolutionparameter. The commit hash encoding and puzzle verification calls properly handle the larger array.
214-229: LGTM!The
getPlantedSecretfunction correctly:
- Returns
int8[128] memory secret- Uses dynamic block calculation
- Includes boundary check for non-multiple-of-16 dimensions
248-324: LGTM!The
_verifyPuzzlefunction correctly implements the LWE verification for n=128, m=192:
- Ternary validation over all 128 elements
- Dynamic block calculation for planted secret and matrix multiplication
- Proper boundary checks with
if (col >= N_WEAK) break- Threshold comparison against
PUZZLE_THRESHOLD_SQ = 800
20-23: Good documentation honesty.The comment correctly:
- States the brute-force floor as 3^128
- Notes "Lattice security TBD via estimator" acknowledging that formal lattice security analysis is pending
- Clarifies that low-entropy secrets remain dictionary-bound without rate limits
This transparency about security claims is appropriate for an educational example.
✏️ Tip: You can disable this entire section by setting review_details to false in your review settings.
|
@coderabbitai resolve |
✅ Actions performedComments resolved. |
…and docs - scripts/BenchmarkTLOS.s.sol: Update puzzle params n=128, m=192 - examples/TLOSSealedAuction.sol: Fix comment int8[64] -> int8[128] - README.md: Fix lattice estimator stddev 8.0 -> 25.0
* feat(security): harden puzzle layer params (n=128, m=192) - Increase Layer 4 puzzle parameters: - Secret dimension: n=48 → n=128 - Samples: m=72 → m=192 - Threshold: 300 → 800 - Search space: 3^48 (~2^76) → 3^128 (~2^203 brute-force) - Update contracts: TLOSWithPuzzleV5.sol, WeakLWEPuzzleV7.sol - Update examples: TLOSDeadManSwitch, TLOSRecovery, TLOSSealedAuction, TLOSTreasureHunt - Update tests with int8[128] arrays and gas threshold assertions - Update docs: AGENTS.md, README.md, docs/layers/layer4-puzzle/ - Add build artifacts to .gitignore Gas: ~8.6M puzzle verification (14% of 60M block limit) All 157 tests passing. Closes #70 * Add Wycheproof-style vectors and test runner * docs: update Layer 4 security docs with GPU benchmark results - Add dictionary attack benchmarks (A100: 5.8M guesses/sec, 0.17µs/guess) - Remove misleading 'unless rate-limited' caveats (doesn't apply offline) - Add explicit attack cost formula: |Dictionary| × 0.17µs (GPU) - Clarify 3^128 floor applies to planted-solution recovery only - Recommend high-entropy secrets for offline security - Add modal_puzzle_benchmark.py for reproducible benchmarks - Add KANBAN.md for tracking Addresses reviewer concern in issue #65 comment 3753688226 Amp-Thread-ID: https://ampcode.com/threads/T-019bc5e7-3901-7776-9a43-4adb6762c244 Co-authored-by: Amp <amp@ampcode.com> * docs: clarify hash_solution comment for Solidity fixed-size array encoding * fix: address PR #74 review nits - update benchmark script, comments, and docs - scripts/BenchmarkTLOS.s.sol: Update puzzle params n=128, m=192 - examples/TLOSSealedAuction.sol: Fix comment int8[64] -> int8[128] - README.md: Fix lattice estimator stddev 8.0 -> 25.0 * fix: update lattice_estimator.rs stddev 8.0 -> 25.0 to match production params PR #69 increased NOISE_SIGMA from 8.0 to 25.0 but the estimator still hardcoded the old value. Fixes AI review nit from PR #69. * chore: stop tracking generated artifacts * docs: note PR description newline rule * docs: align security params and tidy markdown * docs: update Tenderly deploy costs * chore: set benchmark expiry to max * chore: ignore docs plans * chore: keep docs plans untracked --------- Co-authored-by: Igor <igor53627@users.noreply.github.com> Co-authored-by: Amp <amp@ampcode.com>
Summary
Increases Layer 4 puzzle parameters for stronger security:
Changes
contracts/TLOSWithPuzzleV5.solandcontracts/WeakLWEPuzzleV7.solwith new paramsint8[128]arrays and gas threshold assertionsGas Impact
Puzzle verification: ~8.6M gas (14% of 60M block limit)
Testing
All 157 tests passing.
Closes #70
Note
Strengthens the planted LWE puzzle and aligns the codebase and docs with the new parameters and gas figures.
n=128, m=192, q=2039, threshold=800with 3^128 brute-force floor; updatesTLOSWithPuzzleV5verification to loadbas 192×u16 and adds a critical note that noise must be secretint8[128](revealWithPuzzle,checkWithPuzzle,_verifyPuzzle) and propagates across examples (TLOSDeadManSwitch,TLOSRecovery,TLOSSealedAuction,TLOSTreasureHunt)WeakLWEPuzzleV7to hardened params for gas testing only (deterministic noise warning) and adjusts block-based coefficient generationREADME.md,AGENTS.md,docs/layers/layer4-puzzle/) with σ=25 LWE, new puzzle params, and Tenderly gas:checkWithPuzzle()totals and puzzle verification ≈ 8.62M (14% of 60M).gitignoreand regenerates build cache/artifacts (includingIHoneypotmetadata)Written by Cursor Bugbot for commit 87082b9. Configure here.
Summary by CodeRabbit
Documentation
Refactor
Tests
Chores
✏️ Tip: You can customize this high-level summary in your review settings.