CtrlK
BlogDocsLog inGet started
Tessl Logo

dmk-intent-vocabulary

Map developer intent and natural language to the correct Ledger Device Management Kit (DMK) components, operations, and API methods. Use when a developer's request is phrased informally, incompletely, or in terms of outcomes rather than API methods, when the intent is ambiguous and needs mapping to a specific DMK component or operation, or when a developer asks 'how do I...', 'what's the difference between...', or describes a symptom rather than an API.

94

1.15x
Quality

Does it follow best practices?

Impact

98%

1.15x

Average score across 3 eval scenarios

SecuritybySnyk

Advisory

Suggest reviewing before use

SKILL.md
Quality
Evals
Security

Evaluation results

100%

5%

Ethereum Receive Address Widget

ETH receive address with on-device verification

Criteria
Without context
With context

Discovery before connect

100%

100%

Correct connect signature

50%

100%

sessionId used for subsequent calls

100%

100%

OpenAppDeviceAction used

100%

100%

signerEth.getAddress called

100%

100%

checkOnDevice: true

100%

100%

Derivation path as literal

100%

100%

getAddress returns observable

100%

100%

No raw APDU for address

100%

100%

100%

4%

DeFi Signing Flows for a Multi-Protocol Wallet

DeFi multi-protocol Ethereum signing flows

Criteria
Without context
With context

signTransaction method

100%

100%

signTransaction observable handling

100%

100%

signTypedData method

100%

100%

signTypedData full object

100%

100%

signMessage method

100%

100%

eth_sign risk flagged

100%

100%

Derivation paths as literals

66%

100%

No raw APDU for signing

100%

100%

User rejection as error

100%

100%

signMessage safety prefix noted

100%

100%

94%

29%

Ledger DMK Integration Test Suite Setup

CI test setup, custom command, and device-guard utility

Criteria
Without context
With context

Speculos transport used

0%

100%

DevToolsLogger or DevToolsDmkInspector used

0%

25%

Production warning present

100%

100%

No setStub(true) in non-test code

100%

100%

Command class extended

80%

100%

dmk.sendCommand used

0%

100%

DeviceSessionState checked first

100%

100%

Locked state check

100%

100%

No raw sendApdu for custom command

100%

100%

getDeviceSessionState returns observable

100%

100%

Repository
LedgerHQ/agent-skills
Evaluated
Agent
Claude Code
Model
Claude Sonnet 4.6

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.