reference-sdk — Claude Code skill: what it does, how to install
Inspects reference SDK implementations through the btca utility to guide feature development, error handling, and multi-provider code patterns.
npx skills add parcadei/Continuous-Claude-v3 --skill reference-sdk -a claude-codeWhat reference-sdk does
This skill instructs the agent to query reference codebases via btca ask before building new SDK features. It comes preconfigured for three resources: vercel-ai for streaming and structured output, anthropic-sdk for provider-specific API calls, and zod for validation patterns.
The skill runs autonomously with user-invocable: false during debugging or feature design. It strictly relies on the external btca CLI: without this tool and its configured resources, the agent cannot query the upstream repositories. Teams can append more source codebases with btca config resources add -n <name> -t git -u <url> -b <branch>. The instructions explicitly advise against using this skill for docs lookups or project-local searches.
What it helps with
- inspect streaming and structured output patterns in Vercel AI SDK
- check tool definition and calling structures in Anthropic SDK
- examine error handling and validation logic inside Zod safeParse
- compare edge-case handling across different provider implementations
Example prompts for the agent
- How does streamObject work in the Vercel AI SDK? Check the reference code.
- Inspect how the Anthropic SDK handles tool calling and definitions.
- Check how Zod safeParse handles errors in its reference implementation.
How to install reference-sdk
1Skills CLI (Vercel)
npx skills add parcadei/Continuous-Claude-v3 --skill reference-sdk -a claude-code2Manually: put the files in place
~/.claude/skills/reference-sdk/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/reference-sdk/
1Skills CLI (Vercel)
npx skills add parcadei/Continuous-Claude-v3 --skill reference-sdk -a codex2Manually: put the files in place
.agents/skills/reference-sdk/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add parcadei/Continuous-Claude-v3 --skill reference-sdk -a cursor2Manually: put the files in place
.cursor/skills/reference-sdk/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install parcadei/Continuous-Claude-v3 reference-sdk2Manually: put the files in place
.github/skills/reference-sdk/SKILL.md1Skills CLI (Vercel)
npx skills add parcadei/Continuous-Claude-v3 --skill reference-sdk -a gemini-cli2Platform CLI
gemini skills install https://github.com/parcadei/Continuous-Claude-v3 --path .claude/skills/reference-sdkManually: put the files in place
.devin/skills/reference-sdk/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add parcadei/Continuous-Claude-v3 --skill reference-sdk -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about reference-sdk
How to install reference-sdk in Claude Code?
Run npx skills add parcadei/Continuous-Claude-v3 --skill reference-sdk -a claude-code in your project terminal.
What does reference-sdk do?
It directs the coding agent to inspect reference repositories through btca ask to see how established libraries solve similar problems.
When should you avoid using reference-sdk?
Avoid using it for standard documentation lookups, generic web search questions, or searching patterns inside your local project codebase.
Source text
SKILL.md · MIT
--- name: reference-sdk description: Check reference SDK implementations using btca ask user-invocable: false --- # Reference SDK Check When implementing SDK features or debugging provider-specific issues, check reference implementations. ## When to Use - Implementing SDK features - Debugging provider-specific issues - Understanding how other libraries solve similar problems - "How does Vercel AI SDK do X?" - "Check Anthropic SDK for Y" ## Commands Use `btca ask` to check how reference SDKs implement similar features: ```bash # Check Vercel AI SDK for streaming patterns btca ask -r vercel-ai -q "How does streamObject work?" # Check Anthropic SDK for tool calling btca ask -r anthropic-sdk -q "How are tools defined and called?" # Check Zod for validation patterns btca ask -r zod -q "How does safeParse handle errors?" ``` ## Configured Resources - `vercel-ai` - Streaming, tool calling, structured output - `anthropic-sdk` - Anthropic API patterns - `zod` - Schema validation Add more: `btca config resources add -n <name> -t git -u <url> -b <branch>` ## When to Check 1. **Before implementing** - See how others solved similar problems 2. **When debugging** - Find how reference code handles edge cases 3. **Multi-provider support** - Compare implementations across SDKs 4. **Validation patterns** - Check idiomatic approaches ## Don't Use For - Documentation lookups (use /nia-docs instead) - Simple API questions (use WebSearch) - Project-specific patterns (use Grep/Glob)
© parcadei · MIT · parcadei/Continuous-Claude-v3