Hedera Skills

A marketplace of plugins and skills for AI coding agents. Includes Hedera-specific development tools and general-purpose dev workflow intelligence. Each plugin contains packaged instructions, references, and examples that extend agent capabilities.

Installation

Claude Code

# Add the Hedera marketplace
/plugin marketplace add hedera-dev/hedera-skills

# Install individual plugins
/plugin install agent-kit-plugin
/plugin install system-contracts
/plugin install oracles
/plugin install cross-chain
/plugin install native-services-js
/plugin install hackathon-helper
/plugin install hedera-harness
/plugin install dev-intelligence

Other Agents (npx skills)

npx skills add hedera-dev/hedera-skills

Skills are automatically available once installed. The agent will use them when relevant tasks are detected.

Available Plugins

agent-kit-plugin

Comprehensive guide for creating custom plugins that extend the Hedera Agent Kit. Enables developers to add new tools for Hedera network interactions.

Use when:

• Creating a new plugin for Hedera Agent Kit
• Adding custom tools for Hedera network operations
• Learning plugin architecture and patterns
• Building mutation tools (token creation, transfers)
• Building query tools (balance queries, token info)

Topics covered:

• Quick start guide (5-step process)
• Plugin interface documentation
• Tool interface specifications
• Zod schema patterns for Hedera types
• Prompt writing patterns
• Error handling and output parsing
• Working code examples (simple-plugin, token-plugin)

References included:

• plugin-interface.md - Complete interface and type definitions
• zod-schema-patterns.md - Parameter validation patterns for Hedera operations
• prompt-patterns.md - Effective tool description writing
• error-handling.md - Error handling and output parsing patterns

system-contracts

Technical references for Hedera system contracts — the precompiled smart contract APIs for interacting with Hedera native services from Solidity.

Skills included:

• hts-system-contract — Hedera Token Service system contract (0x167). Token creation (fungible and NFT), minting, burning, transfers, association model, key system, fees, and compliance features.
• hss-system-contract — Hedera Schedule Service system contract (0x16b). Scheduling native HTS token creation, generalized scheduled contract calls, and schedule signing from contracts (HIP-755, HIP-756, HIP-1215).

Use when:

• Writing Solidity contracts that interact with HTS or HSS
• Creating or managing tokens via smart contracts
• Scheduling transactions from within contracts
• Understanding HTS response codes and error handling
• Configuring token keys and permissions
• Working with token fees and compliance features

References included (HTS):

• api.md - HTS contract API reference (Solidity signatures)
• structs.md - Data structure definitions (HederaToken, TokenKey, Expiry)
• response-codes.md - Response codes with troubleshooting
• keys.md - Token key system reference
• fees.md - Fee structure information
• compliance.md - Compliance-related features
• troubleshooting.md - Common issues and solutions

References included (HSS):

• api.md - HSS contract API reference (Solidity signatures)

oracles

Hedera price oracle integrations — one skill per provider, matching the cross-chain plugin layout. Skills stay generic (read/update rules, Hedera quirks, where to source IDs). Project-specific adapter names, deploy runbooks, and feed tables belong in the consuming repo’s AGENTS.md.

Skills included:

• chainlink-data-feeds — AggregatorV3 Data Feeds on Hedera (latestRoundData completeness, staleness, decimal normalization). Not CCIP.
• supra-push-oracle — Supra S-Value push (getSvalue), millisecond timestamps, and USDT-quoted pairs on Hedera.
• pyth-price-feeds — Pyth pull feeds: Hermes update payloads, payable updatePriceFeeds, getPriceNoOlderThan, confidence, and signed expo normalization.

Use when:

• Reading Chainlink Data Feeds on Hedera testnet (296) or mainnet (295)
• Integrating Supra push oracles and converting Hedera timestamps to unix seconds
• Pulling Pyth prices via Hermes updates before on-chain reads
• Normalizing provider decimals / exponents to a consumer price scale

References included:

• examples.md - Provider read / update sketches per skill
• hedera-feeds.md - Where to source Hedera feed addresses, Supra pair IDs, and Pyth price IDs

cross-chain

Cross-chain interoperability patterns for Hedera — Axelar GMP, LayerZero V2 OFT, and Chainlink CCIP CCT bridges, with peer/gas/selector wiring, allowlisting, and bridge-agnostic orchestration seams.

Skills included:

• axelar-gmp — Axelar Gateway callContract, gas service payment, AxelarExecutable receivers, Hedera↔EVM wiring, and IBridgeSender / handler split (as used by cross-chain DCA orchestration + HSS).
• layerzero-messaging — LayerZero V2 OFT / OApp on Hedera: Endpoint peers, ULN/DVN/executor config, quoteSend/send, and HTS-backed connector OFTs (mint/burn via 0x167).
• ccip — Chainlink CCIP burn-and-mint Cross-Chain Tokens: chain selectors, TokenAdminRegistry, ccipSend, and Hedera HTS-backed wrappers vs vanilla ERC-20. Not Data Feeds.

Use when:

• Sending cross-chain contract calls from Hedera via Axelar GMP
• Implementing AxelarExecutable receivers on EVM destinations
• Wiring destination/source addresses after dual-chain deploys
• Separating bridge transport from destination business logic
• Combining HSS scheduled execution with Axelar message dispatch
• Building Hedera↔EVM OFT bridges with LayerZero V2
• Configuring setPeer, send/receive libraries, and enforced options
• Implementing HTS connector OFTs (burn on send / mint on receive)
• Bridging burn-and-mint tokens with Chainlink CCIP
• Registering CCIP token pools and wiring remote chain selectors
• Using an HTS-backed CCIP wrapper (users hold native HTS; CCIP sees the wrapper)

References included:

• examples.md - Sender, receiver, orchestrator, executor, OFT, CCIP register/wire/send, and HTS dual-approve skeletons
• hedera-axelar.md - Testnet gateway/gas addresses, Axelar chain names, env vars
• hedera-endpoints.md - LayerZero EIDs, Endpoint V2, ULN, DVN, and executor addresses
• hedera-ccip.md - CCIP selectors vs chain IDs, router/registry lookup, HTS wrapper notes

native-services-js

Comprehensive guides for using Hedera native services with the Hiero JavaScript SDK, plus x402 pay-per-use payment patterns on Hedera.

Skills included:

• hedera-token-service — Token creation (fungible and NFT), minting, burning, transfers, key roles, compliance operations (KYC, freeze, wipe, pause), airdrops, and custom fees using the Hiero JS SDK.
• hedera-consensus-service — Topic creation, message submission with chunking support, subscription patterns via mirror nodes, topic management, and common patterns (event logs, pub/sub).
• x402-payments — x402 HTTP 402 pay-per-use with native HBAR: FileRegistry metadata, self-hosted facilitator verify/settle, and HashPack client payment retries.

Use when:

• Building JavaScript/TypeScript apps that interact with Hedera Token Service
• Creating, minting, or transferring tokens using the Hiero JS SDK
• Working with Hedera Consensus Service topics and messages
• Setting up custom fees, compliance operations, or airdrops
• Subscribing to topic messages via mirror nodes
• Gating downloads or APIs behind x402 HBAR payments on Hedera
• Wiring a self-hosted x402 facilitator or ExactHederaScheme resource server

References included (HTS):

• api-reference.md - Hiero JS SDK API reference for HTS
• custom-fees.md - Custom fee configuration (fixed, fractional, royalty)

References included (HCS):

• api-reference.md - Hiero JS SDK API reference for HCS

References included (x402):

• examples.md - FileRegistry, resource server, and client retry skeletons
• facilitator.md - Facilitator endpoints, fee-payer env vars, Docker infra

hackathon-helper

Two skills for Hedera hackathon participants: project planning and submission validation, both aligned to the official judging criteria. Compatible with any AI coding agent that supports skills (Claude Code, Codex, Gemini CLI, etc.).

Skills included:

• hackathon-prd - Interactive PRD generator. Asks participants to paste their bounty/track context, gathers project details, then generates a comprehensive PRD (HACKATHON-PRD.md) with a predicted score and improvement tips.
• validate-submission - Codebase reviewer. Scans the repo for Hedera integration depth, code quality, and documentation, then produces a weighted scorecard against all 7 judging criteria with prioritized action items.

Use when:

• Starting a Hedera hackathon project and need a structured plan
• Want to ensure your project addresses all judging criteria
• Ready to validate your submission before the deadline
• Need to identify quick wins to improve your hackathon score

Judging criteria covered:

• Innovation (10%) - Novelty in the Hedera ecosystem
• Feasibility (10%) - Web3 necessity, business model, domain knowledge
• Execution (20%) - MVP quality, code quality, UI/UX, strategy
• Integration (15%) - Hedera service depth, ecosystem partners, creativity
• Validation (15%) - Market feedback, early adopters, traction
• Success (20%) - Hedera account growth, TPS impact, audience exposure
• Pitch (10%) - Problem/solution clarity, metrics, Hedera representation

hedera-harness

Three skills for creating and reviewing hedera-harness recipes — the PRD, recipe file, validators, Playwright smoke, and evaluate checklist that drive in-place Scaffold HBAR feature work. Schema v3. Works in Claude Code (marketplace plugin) and Cursor / other skill-capable agents.

Skills included:

• harness-spec-anatomy — Shared vocabulary (recipe / slug / blind / evaluate checklist / stage / needle). Single source of truth for file layout and the mechanical check-spec.sh script.
• create-harness-spec — Grills a product idea one question at a time, then emits an ASSERT recipe (optional SMOKE / EVALUATE / CHAIN). Prefers Matt Pocock /grilling when available; Claude Code uses the inline protocol.
• review-harness-spec — Two-axis audit (Wiring via check-spec.sh + Eval judgment) before a run.

Use when:

• Turning a Hedera feature idea into hedera-harness inputs
• Writing a harness PRD, recipe file, or evaluate checklist
• Reviewing a harness recipe before hedera-harness run
• Deciding which validation stages to enable and in what order
• Splitting a feature into ordered prd: / eval: increments

Important: These are authoring skills. Recipes do not list skills — product plugins are discovered per run and the generator picks. Keep authoring skills as marketplace plugins.

dev-intelligence

AI development workflow toolkit — session continuity, quality gates, project scaffolding, and tech debt tracking for any codebase. Works with any language or framework.

Skills included:

• session-management — Report registry (15 categories), tech debt tracker (P0-P3 priorities), archive strategy for keeping session state lean across conversations.
• quality-gates — PostToolUse auto-validation hooks for TypeScript, Python, Rust, and Go. Deploy checklists and local CI pipeline patterns.
• project-scaffolding — Generate CLAUDE.md and .claude/ directory structure for any project. Auto-detects stack, naming conventions, and build commands.

Commands included:

• /continue — Resume work with full context (git log, registry, tech debt)
• /init — Scaffold a project for AI development (interactive setup)
• /debt — View and manage tech debt tracker
• /health — Run project health check (git, tests, lint, dependencies)

Use when:

• Starting a new project and want AI-ready scaffolding
• Resuming work and need to pick up where you left off
• Want automatic validation after every file edit
• Tracking tech debt across sessions
• Running pre-deploy or pre-push quality checks

Hooks included:

• PostToolUse (Edit/Write) — Auto-runs stack-appropriate linter/type-checker after every edit

Marketplace Structure

hedera-skills/
├── .claude-plugin/
│   └── marketplace.json      # Marketplace manifest
├── plugins/
│   ├── agent-kit-plugin/     # Agent Kit plugin development
│   │   └── skills/
│   │       └── agent-kit-plugin/
│   │           ├── SKILL.md
│   │           ├── examples/
│   │           └── references/
│   ├── hackathon-helper/     # Hackathon PRD & validation
│   │   └── skills/
│   │       ├── hackathon-prd/
│   │       │   ├── SKILL.md
│   │       │   └── references/
│   │       └── validate-submission/
│   │           ├── SKILL.md
│   │           └── references/
│   ├── system-contracts/     # Hedera system contract references
│   │   └── skills/
│   │       ├── hts-system-contract/
│   │       │   ├── SKILL.md
│   │       │   └── references/
│   │       └── hss-system-contract/
│   │           ├── SKILL.md
│   │           └── references/
│   ├── oracles/              # Price oracles (Chainlink, Supra, Pyth)
│   │   └── skills/
│   │       ├── chainlink-data-feeds/
│   │       │   ├── SKILL.md
│   │       │   └── references/
│   │       ├── supra-push-oracle/
│   │       │   ├── SKILL.md
│   │       │   └── references/
│   │       └── pyth-price-feeds/
│   │           ├── SKILL.md
│   │           └── references/
│   ├── cross-chain/          # Cross-chain interoperability (Axelar, LayerZero)
│   │   └── skills/
│   │       ├── axelar-gmp/
│   │       │   ├── SKILL.md
│   │       │   └── references/
│   │       ├── layerzero-messaging/
│   │       │   ├── SKILL.md
│   │       │   └── references/
│   │       └── ccip/
│   │           ├── SKILL.md
│   │           └── references/
│   ├── native-services-js/   # Hedera native services + x402 payments
│   │   └── skills/
│   │       ├── hedera-token-service/
│   │       │   ├── SKILL.md
│   │       │   └── references/
│   │       ├── hedera-consensus-service/
│   │       │   ├── SKILL.md
│   │       │   └── references/
│   │       └── x402-payments/
│   │           ├── SKILL.md
│   │           └── references/
│   ├── hedera-harness/       # Harness spec authoring & review
│   │   └── skills/
│   │       ├── harness-spec-anatomy/
│   │       │   ├── SKILL.md
│   │       │   ├── GLOSSARY.md
│   │       │   ├── scripts/
│   │       │   └── references/
│   │       ├── create-harness-spec/
│   │       │   ├── SKILL.md
│   │       │   ├── evals/
│   │       │   └── references/
│   │       └── review-harness-spec/
│   │           ├── SKILL.md
│   │           ├── evals/
│   │           └── references/
│   └── dev-intelligence/     # Dev workflow intelligence
│       ├── skills/
│       │   ├── session-management/
│       │   │   ├── SKILL.md
│       │   │   └── references/
│       │   ├── quality-gates/
│       │   │   ├── SKILL.md
│       │   │   └── references/
│       │   └── project-scaffolding/
│       │       ├── SKILL.md
│       │       └── references/
│       ├── commands/
│       ├── hooks/
│       └── scripts/
└── README.md

Each plugin contains:

• skills/<name>/SKILL.md - Instructions for the agent
• skills/<name>/references/ - Supporting documentation
• skills/<name>/examples/ - Working code examples (optional)
• skills/<name>/evals/spec.json - Structured eval checks (source of truth)
• skills/<name>/evals/evals.json - Generated assertions for agent-skills-eval

Skill Evaluations

Skills with automated evals use a two-file layout:

File	Purpose
evals/spec.json	Structured checks (name, type, value, description, optional rubric) — edit this
evals/evals.json	Generated string assertions for agent-skills-eval — do not edit by hand

Compile before running evals:

npm run evals:compile

Verify generated files are up to date (for CI):

npm run evals:compile:check

Run evals for a skill (requires OPENAI_API_KEY and optional OPENAI_BASE_URL for OpenRouter):

set -a && source .env && set +a

npx agent-skills-eval ./plugins/native-services-js/skills/hedera-token-service \
  --target deepseek/deepseek-v4-flash \
  --judge deepseek/deepseek-v4-flash \
  --baseline \
  --report

Add optional rubric on a check when the auto-generated judge text is too brittle (for example regex-based checks).

License

Apache-2.0