Add add-chain Claude skill for adding chains/tokens

[reednaa]

What

Adds a self-contained Claude Code skill (.claude/skills/add-chain/SKILL.md) that lets an agent add a new EVM chain and/or tokens to the demo app in one focused turn, for demoing the service on new networks.

The skill edits only src/lib/config.ts (the single source of truth for chains — wallet onboarding auto-derives from chainMap) across its 7 coupled sites, and handles:

• viem-supported chains (wraps the export so the user's display name + custom RPC reach chainMap) and non-viem chains (defineChain).
• Custom RPCs (prepended as primary, viem defaults as fallback).
• Native + ERC-20 tokens — fetches decimals/symbol via RPC; native uses ADDRESS_ZERO.
• Demo-readiness verification (non-blocking): chainId, infra codehash compare vs a reference chain, Polymer info_health, RouteMesh, CORS — plus bun run check as the one blocking gate.
• Ends with a usage + verification summary. Does not commit or open a PR itself.

Also narrows the .claude gitignore so the skill is tracked:

.claude/*
!.claude/skills/

(local settings/worktrees stay ignored).

How it was built

Design and the skill file were each independently reviewed (review subagent + Codex oracle); findings on secret handling, the viem-wrap requirement, native-token templating, verification rigor, and the duplicate-chain guard were folded in.

Validation

Smoke-tested end-to-end on a viem chain (Optimism + USDC + native ETH): all 7 edits applied, invariant check passed, bun run check introduced zero new errors (verified against baseline), infra codehashes matched base, and Polymer reported chain 10 healthy. Reverted afterward — this PR contains only the skill + gitignore change.

Note: bun run check currently fails on main for an unrelated reason — local .env lacks PRIVATE_ROUTEMESH_API_KEY (present in .env.example). The skill documents this baseline gotcha.

🤖 Generated with Claude Code

Summary by CodeRabbit

• Documentation

  • Added a detailed guide for adding new EVM chains and tokens to the application, including step-by-step procedures, validation requirements, and verification strategies.

• Chores

  • Updated Git configuration to manage skill documentation files.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 3f1eb5c5-a36f-433f-b7d3-29f4b8446c1a

📥 Commits

Reviewing files that changed from the base of the PR and between fd3980c and 4f74f1e.

📒 Files selected for processing (2)

• .claude/skills/add-chain/SKILL.md
• .gitignore

---

📝 Walkthrough

Walkthrough

Adds .claude/skills/add-chain/SKILL.md, a 332-line procedural skill document for an AI assistant describing how to add EVM chains and tokens to src/lib/config.ts. Updates .gitignore to ignore .claude/* broadly while explicitly unignoring .claude/skills/ so the new file is tracked.

Changes

Add-chain skill document and gitignore wiring

Layer / File(s)	Summary
Gitignore exposure and skill front-matter/rules .gitignore, .claude/skills/add-chain/SKILL.md	.gitignore switches from ignoring .claude as a whole to ignoring .claude/* with a negation for .claude/skills/. SKILL.md opens with name: add-chain metadata, hard rules restricting edits to src/lib/config.ts, a once-only question policy, and bun run check as the sole blocking gate.
Chain/token resolution logic and 7-edits checklist .claude/skills/add-chain/SKILL.md	Steps 1–3: chainId lookup, viem canonicalization, mainnet/testnet bucketing, native vs ERC-20 token metadata fetching, naming collision rules, and a checklist of seven required edits to src/lib/config.ts (chain objects, chainMap, chain lists, POLYMER_ORACLE, coinList, clients, polymerChainIds).
Invariant checks, verification, output template, and appendix .claude/skills/add-chain/SKILL.md	Steps 4–6 and appendix: deterministic invariant check, bun run check compile gate, non-blocking external verifications (infra codehash comparisons, Polymer health, optional RouteMesh, CORS), a structured deliverable output template with a ✅/⚠️/❌ verification table, and a reference appendix of infra constants, RPC endpoints, and tooling.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Poem

🐇 A skill was born in a hidden lair,
.claude/skills/ now floats in the air.
Seven edits, a bun run check,
No commits made — what the heck!
The rabbit hops from chain to chain,
Tokens added, nothing in vain. ✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title directly summarizes the main change: adding a Claude skill for adding chains and tokens to the demo app, which aligns perfectly with the PR's core objective.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch add-chain-skill

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[github-actions]

🚀 Preview deployed!

Worker: lintent-pr-55
URL: https://lintent-pr-55.li-fi374.workers.dev