TIG Innovator Skill

Autonomous algorithm optimization agent for The Innovation Game (TIG) protocol. This skill enables AI agents to earn cryptocurrency by optimizing algorithms and submitting improvements to TIG's decentralized marketplace.

Overview

TIG is a protocol that incentivizes algorithmic innovation. Innovators submit algorithm optimizations that compete for adoption by benchmarkers. If your algorithm is adopted by 25%+ of benchmarkers, you earn a share of block rewards.

How this skill makes money:

Download existing TIG algorithms (free)
Analyze code with LLM → suggest Rust optimizations
Test locally to verify improvement (free)
Submit only when competitive (costs 10 TIG ≈ $1)
Earn recurring rewards if adopted

Ecosystem Integration

This skill can be wrapped as a paid Lucid Agent service -- offering algorithm optimization-as-a-service via x402 micropayments. Use lucid-agents-sdk to expose TIG optimization as paid entrypoints (e.g., analyze, optimize, benchmark).

Quick Start

# 1. Set up environment
export ANTHROPIC_API_KEY="your-key"
export TIG_NETWORK="testnet"  # Start with testnet

# 2. List algorithms for a challenge
tig-innovator list vector_search

# 3. Analyze an algorithm for optimization opportunities
tig-innovator analyze <algorithm-id>

# 4. Generate optimized version
tig-innovator optimize <algorithm-id>

# 5. Test the optimization locally
tig-innovator test <algorithm-id> --optimized ./optimized.rs

# 6. Submit if improvement is significant (>5%)
tig-innovator submit <algorithm-id> --code ./optimized.rs

Commands

`tig-innovator list <challenge>`

Lists all algorithms for a challenge with performance metrics.

Arguments:

challenge: One of satisfiability, vehicle_routing, knapsack, vector_search, hypergraph, neural_network

Options:

--network <mainnet|testnet>: Network to query (default: mainnet)
--limit <n>: Max algorithms to show (default: 20)
--sort <adoption|score>: Sort order (default: adoption)

Example:

tig-innovator list vector_search --limit 10 --sort adoption

`tig-innovator analyze <algorithm-id>`

Downloads and analyzes an algorithm's source code for optimization opportunities.

Arguments:

algorithm-id: The algorithm ID from the list command

Options:

--detailed: Show detailed analysis with code snippets

Output:

Optimization opportunities ranked by impact
Specific code locations
Suggested improvements
Confidence scores

Example:

tig-innovator analyze c003_a042_0001

`tig-innovator optimize <algorithm-id>`

Generates an optimized version of the algorithm using LLM analysis.

Arguments:

algorithm-id: The algorithm ID to optimize

Options:

--output <path>: Output path for optimized code (default: ./optimized_.rs)
--iterations <n>: Number of optimization iterations (default: 3)
--focus <type>: Focus on specific optimization type: loop, memory, simd, cache, all

Example:

tig-innovator optimize c003_a042_0001 --output ./my_optimization.rs --iterations 5

`tig-innovator test <algorithm-id>`

Runs local benchmarks comparing original vs optimized algorithm.

Arguments:

algorithm-id: The baseline algorithm ID

Options:

--optimized <path>: Path to optimized code file
--samples <n>: Number of test samples (default: 100)

Requirements:

Docker must be running (uses TIG dev containers)

Example:

tig-innovator test c003_a042_0001 --optimized ./my_optimization.rs

`tig-innovator submit <algorithm-id>`

Submits optimized algorithm to TIG network.

Arguments:

algorithm-id: Reference to the algorithm being improved

Options:

--code <path>: Path to optimized code file
--network <mainnet|testnet>: Target network (default: from env)
--dry-run: Validate without submitting

Requirements (mainnet only):

TIG_WALLET_ADDRESS environment variable
TIG_PRIVATE_KEY environment variable
Sufficient TIG balance (10 TIG per submission)

Example:

# Testnet (free)
tig-innovator submit c003_a042_0001 --code ./optimized.rs --network testnet

# Mainnet (costs 10 TIG)
tig-innovator submit c003_a042_0001 --code ./optimized.rs --network mainnet

`tig-innovator status`

Shows status of your submissions and earnings.

Options:

--network <mainnet|testnet>: Network to query

Workflow Recommendations

For Best Results

Start with vector_search or neural_network - These challenges have the most opportunity for micro-optimizations that LLMs excel at (loop unrolling, SIMD, cache optimization).
Always test before submitting - The test command runs real benchmarks. Only submit if you see >5% improvement.
Use testnet first - Validate your workflow on testnet before spending TIG on mainnet.
Focus on adopted algorithms - Algorithms with high adoption have proven benchmarker interest. Improving them increases your chances of adoption.
Iterate multiple times - Run optimize with --iterations 5 or more. Each pass finds new opportunities.

Optimization Types

Type	Description	Best For
`loop`	Loop unrolling, vectorization	All challenges
`memory`	Memory access patterns, allocation	Large data sets
`simd`	SIMD intrinsics	Numeric computation
`cache`	Cache-friendly data layout	Matrix operations
`branch`	Branch prediction hints	Conditional logic

Example Full Workflow

# 1. Find a good target
tig-innovator list vector_search --sort adoption --limit 5

# 2. Analyze top algorithm
tig-innovator analyze c003_a042_0001 --detailed

# 3. Generate optimization (multiple iterations)
tig-innovator optimize c003_a042_0001 \
  --output ./vs_optimized.rs \
  --iterations 5 \
  --focus all

# 4. Test locally
tig-innovator test c003_a042_0001 --optimized ./vs_optimized.rs

# Output shows:
# Baseline score: 1000
# Optimized score: 1120
# Improvement: +12%
# Recommendation: SUBMIT ✓

# 5. Submit to testnet first
tig-innovator submit c003_a042_0001 \
  --code ./vs_optimized.rs \
  --network testnet

# 6. If testnet succeeds, submit to mainnet
tig-innovator submit c003_a042_0001 \
  --code ./vs_optimized.rs \
  --network mainnet

# 7. Monitor adoption
tig-innovator status --network mainnet

Environment Variables

Variable	Required	Description
`ANTHROPIC_API_KEY`	Yes	API key for Claude (code analysis)
`TIG_NETWORK`	No	Default network: `mainnet` or `testnet`
`TIG_WALLET_ADDRESS`	Mainnet	Your TIG wallet address
`TIG_PRIVATE_KEY`	Mainnet	Wallet private key for submissions
`TIG_API_URL`	No	Custom API URL (default: api.tig.foundation)

Challenges Reference

Challenge	Description	Optimization Potential
`vector_search`	Nearest neighbor search	HIGH - SIMD, cache
`neural_network`	NN training optimization	HIGH - Matrix ops
`knapsack`	Quadratic knapsack problem	MEDIUM - Loop opt
`satisfiability`	Boolean SAT solving	MEDIUM - Branch pred
`vehicle_routing`	VRPTW optimization	MEDIUM - Memory
`hypergraph`	Graph partitioning	LOW - Algorithm-heavy

Troubleshooting

"Docker not found"

The test command requires Docker. Install Docker and ensure it's running.

"Compilation failed"

The optimized code must compile as valid Rust. Check:

All imports are valid (no external crates allowed)
Function signature matches solve_challenge
Code is deterministic

"Score lower than baseline"

Not all optimizations improve performance. Try:

Different --focus types
More --iterations
Different algorithm as baseline

"Insufficient TIG balance"

Mainnet submissions require 10 TIG. Get TIG from:

Exchanges (if listed)
TIG faucet (testnet only)
Delegation rewards

Architecture

tig-innovator/
├── src/
│   ├── cli.ts              # CLI entrypoint
│   ├── tig/
│   │   ├── api.ts          # TIG API client
│   │   ├── algorithms.ts   # Algorithm fetching
│   │   └── submission.ts   # Submission logic
│   ├── optimizer/
│   │   ├── analyzer.ts     # LLM code analysis
│   │   ├── generator.ts    # Optimization generation
│   │   └── validator.ts    # Code validation
│   └── benchmark/
│       ├── runner.ts       # Docker benchmark runner
│       └── comparator.ts   # Score comparison
└── docker/
    └── tig-dev.dockerfile  # TIG dev environment

License

MIT - Built for the Daydreams AI ecosystem.