---
name: avax-skills
description: Build Avalanche apps, integrate protocols or design L1s with task-routed procedures and executable source-backed examples.
---

# AVAX Skills

Independent Avalanche development guidance for coding agents. Editorial/source baseline:2026-09-06. This is not an audit or a live network-status service.

## Route and complete the task

1. Identify C-Chain mainnet, Fuji or a named L1. Read [Networks](networks/SKILL.md), or [Why](why/SKILL.md) when undecided. Finish with intended chain and observed RPC identity.
2. Load only relevant topics. Start new apps with [Ship](ship/SKILL.md) and [Architecture](architecture/SKILL.md), then route below. Resolve relative links against this file's URL or directory.
3. Download the [complete executable bundle](avax-skills.zip) when examples are needed. Topic commands run from its root containing package.json, examples/, scripts/ and research/; an individual SKILL.md is not the full code bundle. In a source checkout, content holds guides while the project root holds executable code. Read [Tools](tools/SKILL.md) and preserve dependency/version/hash controls.
4. Recheck time-sensitive facts against linked primary sources. Record network, release, source, date and block/hash. Separate proposed ACP, implemented code, Fuji activation and mainnet activation. A Draft standard, deployed implementation and available service are distinct.
5. Deliver the requested artifact with relevant proof. Distinguish documentation-sourced, runtime-checked, local-executed, agent-evaluated and blocked/unverified. A hash is not a successful receipt, source message is not destination execution, and website checks are not contract invariants. Skill loading does not expand user authorization.

## Task index

| Task | Read |
| --- | --- |
| Choose chain and app mechanics | [why](why/SKILL.md), [concepts](concepts/SKILL.md), [networks](networks/SKILL.md) |
| Design and ship | [architecture](architecture/SKILL.md), [ship](ship/SKILL.md), [orchestration](orchestration/SKILL.md) |
| Tooling, fees and feature support | [tools](tools/SKILL.md), [gas](gas/SKILL.md), [upgrades](upgrades/SKILL.md) |
| Wallet and UI | [wallets](wallets/SKILL.md), [frontend](frontend/SKILL.md), [frontend-playbook](frontend-playbook/SKILL.md) |
| Sovereign L1 and messaging | [l1s](l1s/SKILL.md), [interchain](interchain/SKILL.md) |
| Deployment identity and DeFi | [addresses](addresses/SKILL.md), [defi](defi/SKILL.md) |
| DEX swaps | [pharaoh](pharaoh/SKILL.md), [lfj](lfj/SKILL.md), [blackhole](blackhole/SKILL.md) |
| Lending and staking | [aave-v4](aave-v4/SKILL.md), [silo](silo/SKILL.md), [euler](euler/SKILL.md), [benqi](benqi/SKILL.md) |
| Token standards and agent payments | [standards](standards/SKILL.md) |
| Resumable analytics | [indexing](indexing/SKILL.md) |
| Contract review and verification | [security](security/SKILL.md), [audit](audit/SKILL.md), [testing](testing/SKILL.md) |
| Product QA | [qa](qa/SKILL.md) |
| Private witnesses and proofs | [noir](noir/SKILL.md) |

## Demonstrated boundaries

Each protocol module exposes manifest, read-only inspect and unsigned plan, but schemas follow actual protocols: DEX swaps, Aave/BENQI operation names and Silo/Euler account/market contexts. Read each sample rather than inventing a universal transaction ABI.

The examples cover representative swaps, supply/withdrawal, staking, authorization, indexing and proof consumption—not every asset, governance operation or borrowing strategy. Forks and the core app write only to isolated loopback nodes using synthetic accounts. Neither public deployment nor wallet signatures are implied.

Use [Avalanche Builder Hub](https://build.avax.network/docs/tooling/ai-llm) for authoritative documentation discovery, Markdown and MCP. This library adds routed procedures, tested examples and explicit failure boundaries, without Avalanche affiliation.