Skip to main content

AI Skills

The Intuition Agent Skills repository gives AI coding agents canonical protocol context. Use it when you want an agent to produce correct Intuition reads, writes, calldata, or unsigned transaction parameters.

What the Skills Solve​

General LLMs can miss Intuition-specific details, including:

  • L3 chain and deployment details that are not discoverable through Etherscan.
  • V2 contract interfaces and ABIs.
  • bytes32 atom, triple, and term IDs.
  • Batch-only creation flows.
  • Bonding curve and value calculation requirements.

The skills provide verified references so the agent can reason from canonical protocol context instead of guessing.

Install​

Install all Intuition skills:

npx skills add 0xintuition/agent-skills

Install only the core Intuition protocol skill:

npx skills add 0xintuition/agent-skills --skill intuition

Once installed, call the skill from within your agent session:

/intuition

Agents can also invoke the skill autonomously when their runtime supports skills, allowing Hermes-style agents and other AI features to use Intuition context without a manual slash command.

Current Skill​

SkillPurposeRepository
intuitionCanonical reference for Intuition Protocol transactions, ABIs, encoding, addresses, and value calculations.agent-skills

The repository is structured to support more skills over time, but intuition is the current public protocol skill.

Important Boundary​

The skills produce unsigned transaction parameters. Your application, wallet, or backend signing flow remains responsible for:

  • Choosing the signer.
  • Presenting the transaction to the user or signing infrastructure.
  • Broadcasting the transaction.
  • Confirming and handling receipt data.

Treat main as a moving branch. For production agent workflows, pin installs to a Git tag or commit SHA once a release is selected. You can find the latest version and release information in the GitHub releases page.

When to Use AI Skills vs MCP​

Use AI skills when an agent is writing or modifying code that interacts with the protocol.

Use the MCP Server when an AI application needs tools for querying atoms, accounts, social graph data, or lists at runtime.

The two can be used together: skills help the coding agent implement correctly while MCP gives the finished AI application live graph tools.

Pair With Templates​

The official templates already include agent-readable READMEs and .agents/INSTRUCTIONS.md files. Install the Intuition skill before asking an agent to extend those templates with new protocol operations.