agentleFS
Sign inStart free
yu-iskwyu-iskw/skill-inspector/.claude/skills/update-llm-models/SKILL.md

update-llm-models

Skill2 starsChanged 10 months agoApache-2.0View on GitHub

Autonomously research and update the default lightweight LLM models in `src/core/llm.ts` using web search to ensure the most current versions are used.

The file

We split this skill into the rules it gives an agent

12 rules. About 0 are common, 6 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use web_search to identify the most recent lightweight model identifiers for the following providers:

    line 16
  2. Distinctive:

    OpenAI: Look for "mini" or "nano" variants of the latest GPT model (e.g., GPT-5.2).

    line 18
  3. Distinctive:

    Anthropic: Look for "haiku" variants (e.g., Claude 4.5 Haiku).

    line 19
  4. Distinctive:

    Google: Look for "flash" variants (e.g., Gemini 3 Flash).

    line 20
  5. Distinctive:

    Mistral: Look for "small" or "mini" variants (e.g., Mistral Small 3.1).

    line 21
  6. Distinctive:

    Groq: Look for the most efficient models available on Groq (usually Llama 8B or 70B variants).

    line 22
  7. Very specific:

    Update the return values in the switch statement for each provider.

    line 30
  8. Very specific:

    Run pnpm lint to ensure no syntax errors or linting violations.

    line 34
  9. Very specific:

    Run pnpm build to confirm the project still compiles.

    line 35
  10. Very specific:

    Prefer Efficiency: Always choose the "lighter" or "faster" version if multiple variants exist (e.g., prefer gemini-2.5-flash over …

    line 39
  11. Distinctive:

    Exact Identifiers: Use the precise model identifier string required by the provider's API.

    line 40
  12. Very specific:

    Provider Coverage: Ensure all providers in the LLMProvider type are addressed if they have a known lightweight default.

    line 41

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Update LLM Models

Purpose

This skill maintains the src/core/llm.ts file by ensuring that the default model identifiers for each provider are the latest available "lightweight" versions (e.g., Flash, Haiku, Mini).

Workflow

1. Research Latest Models

Use web_search to identify the most recent lightweight model identifiers for the following providers:

  • OpenAI: Look for "mini" or "nano" variants of the latest GPT model (e.g., GPT-5.2).
  • Anthropic: Look for "haiku" variants (e.g., Claude 4.5 Haiku).
  • Google: Look for "flash" variants (e.g., Gemini 3 Flash).
  • Mistral: Look for "small" or "mini" variants (e.g., Mistral Small 3.1).
  • Groq: Look for the most efficient models available on Groq (usually Llama 8B or 70B variants).

2. Identify Target Function

Locate the getDefaultModel function in [src/core/llm.ts](src/core/llm.ts).

3. Apply Updates

Update the return values in the switch statement for each provider. Ensure the model identifiers match the exact strings found during research.

4. Verification

  • Run pnpm lint to ensure no syntax errors or linting violations.
  • Run pnpm build to confirm the project still compiles.

Guidelines

  • Prefer Efficiency: Always choose the "lighter" or "faster" version if multiple variants exist (e.g., prefer gemini-2.5-flash over gemini-2.5-pro).
  • Exact Identifiers: Use the precise model identifier string required by the provider's API.
  • Provider Coverage: Ensure all providers in the LLMProvider type are addressed if they have a known lightweight default.

Examples

Before

case "google":
  return "gemini-2.5-flash";

After (Hypothetical Jan 2026)

case "google":
  return "gemini-3-flash";

Resources

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free