feat: improve developing-genkit-js skill score (86% → 100%)

[yogesh-tessl]

Hey @ssbushi 👋

ran your skills through tessl skill review at work and found some targeted improvements for developing-genkit-js. Here's the before/after:

Skill	Before	After	Change
developing-genkit-js	86%	100%	+14%

Changes made to developing-genkit-js

• Description: Expanded with specific concrete actions (create flows, define tools, configure model plugins, debug pipeline errors) and qualified trigger terms to reduce overlap with generic TypeScript skills
• Core Features table: Added a reference lookup table (matching the pattern from your excellent Go skill) so the agent can quickly navigate to the right reference file based on the task
• Consolidated error handling: The mandate to read common-errors.md first appeared in three separate sections — consolidated into a single authoritative Error Troubleshooting Protocol with brief cross-references
• CLI section: Added concrete key commands inline (start dev server, run flows, search docs) instead of only linking to the reference
• One-liner intro: Added a concise SDK summary line below the heading for immediate context
• Trimmed redundancy: Removed repeated phrasing across "Critical", "Error Troubleshooting Protocol", and "Development Workflow" sections while preserving all guidance

---

also stress-tested your developing-genkit-js skill against a few real-world task evals and it held up really well on multi-provider flow configuration with schema validation. Kudos for that.

quick honest disclosure. I work at https://github.com/tesslio where we build tooling around skills like these. Not a pitch, just saw room for improvement and wanted to contribute.

If you want to self-improve your skills, or define your own scenarios to pressure test, just ask your agent (Claude Code, Codex, etc.) to evaluate and optimize your skill with Tessl. Ping me @yogesh-tessl, if you hit any snags.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[gemini-code-assist]

Code Review

This pull request updates the Genkit JS skill documentation by refining the project description, adding a core features reference table, and streamlining the troubleshooting and development workflows. It also introduces a new section for Genkit CLI usage. Feedback was provided to correct the CLI command from npx genkit to npx genkit-cli and to include the --watch flag in the genkit start examples to ensure the Developer UI automatically reflects code changes.

[gemini-code-assist]

The genkit start command should include the --watch flag to ensure the Developer UI automatically reflects code changes, as recommended in the Best Practices (line 31) and the Setup Guide (line 39). When running via a package manager, ensure you use npx genkit-cli instead of npx genkit.

Suggested change

	genkit start -- npx tsx src/index.ts
	genkit start -o -- npx tsx src/index.ts # also opens browser
	npx genkit-cli start -- npx tsx --watch src/index.ts
	npx genkit-cli start -o -- npx tsx --watch src/index.ts # also opens browser

References

1. The genkit start command should include the --watch flag to ensure the Developer UI automatically reflects code changes. ^(link)
2. When running the Genkit CLI via a package manager like npx, the command is npx genkit-cli, not npx genkit.

[ssbushi]

+1 to using npx, if genkit CLI is not available locally.

[ssbushi]

Hi @yogesh-tessl,

This is very interesting. Thank you for your contribution. The changes in this PR are minor modifications that enchange the wording and clarity. I see nothing jarring here, but I'd like to learn more about the results you got (before and after modifications). Is it possible for you to share more details about the evals process, the test-cases used, etc?

[yogesh-tessl]

@ssbushi, thanks for the kind words and the question. Happy to share more.

How the review scoring works

The scores come from tessl skill review, which runs two checks:

1. Activation score - LLM-as-judge review of the description (specificity, completeness, trigger term quality, distinctiveness)
2. Implementation score - LLM-as-judge review of the SKILL.md body (conciseness, actionability, workflow clarity, progressive disclosure)

Full rubric is documented here: Review a skill against best practices

You can reproduce it locally:

# Install
curl -fsSL https://tessl.io/install.sh | bash
tessl login

# Run against the skill before the change
git checkout main
tessl skill review skills/developing-genkit-js

# Run against the PR branch and compare
git checkout improve/skill-review-optimization
tessl skill review skills/developing-genkit-js

The CLI is free, nothing vendor-locked. I'll add the exact version and raw before/after output to the PR description so you can verify without running it yourself.

Task evals (the scenario-based testing)

Separate from the review score, I ran scenario-based evals where an agent attempts real tasks with and without the skill injected. The process is documented here: Evaluate skill quality using scenarios

The eval results for this skill are public: https://tessl.io/registry/skills/github/genkit-ai/skills/developing-genkit-js/evals

[ssbushi]

Thanks, mostly looks good. Could you please complete the CLA agreement to unblock the checks?

[ssbushi]

I understand the sentiment here, but I feel it is better to focus on the "AI-powered apps/features" terminology. Users may not know that they need to write a flow, but do know they want an AI-powered feature that does xyz.

Suggested change

	description: "Create flows, define tools, configure model plugins, and debug pipeline errors for Genkit in Node.js/TypeScript. Use when building AI agents or flows in JavaScript/TypeScript with Genkit, integrating model providers, defining schemas, or troubleshooting Genkit-specific validation, type, or API errors."
	description: "Develop AI-powered applications using Genkit in Node.js/TypeScript. Use when building AI-powered features, AI agents or workflows in JavaScript/TypeScript with Genkit, integrating model providers, defining schemas, or troubleshooting Genkit-specific validation, type, or API errors."

[ssbushi]

+1 to using npx, if genkit CLI is not available locally.

[yogesh-tessl]

@ssbushi I've addressed all the comments in the new commit and CLA signing is in progress

[yogesh-tessl]

@ssbushi Our Head of Engineering has already signed the CLA on behalf of Tessl, our Employer. However, I can still see that the CLA has not been signed.

I have attached a screenshot for the reference. Please assist me if I am missing out on anything. Thanks!