dmmulroy/better-result/adopt-better-result

Trust tier 2
  • community
LLM review pending
Source check pending

Checking that the SKILL.md file is still in the repository.

Adopt better-result in an existing TypeScript codebase. Use for a repository-wide error-handling audit and proposal, or for implementing a named vertical slice with TaggedError and Result.

Install

npx vskill@latest install dmmulroy/better-result/adopt-better-result

SKILL.md

Adopt better-result

Start with evidence, then work at the scale named by the user.

Choose the branch

  • Audit branch: Use when the user requests a repository-wide audit or proposal, or has not named an implementation slice. Complete the audit branch before changing production code.
  • Named-slice branch: Use when the user explicitly requests implementation of a function, module, boundary, or previously proposed slice. Limit evidence collection to that slice and its propagation path.

An approved slice from the audit branch continues through the named-slice branch.

Audit branch: repository-wide proposal

1. Establish the repository and API facts

Read the repository instructions, package manifests, source layout, test commands, and architecture documentation. Determine whether better-result is installed and which version/API is available. Prefer installed types and source over remembered APIs.

Map every production-code area before studying individual failures. Exclude generated and vendored code from the audit; inspect tests for intended behavior.

Complete when: every production-code area is listed, the validation commands are known, and the available better-result API has been verified.

2. Audit every production-code area

Follow `references/repository-audit.md`. Trace search evidence into end-to-end failure paths, keep grouped sites accountable with counts, and record obscured paths as unknown.

Write the audit to the repository's established documentation or planning location, falling back to docs/better-result-adoption.md.

Complete when: the report satisfies the coverage and completion requirements in repository-audit.md.

3. Design the target model

Apply the failure taxonomy in `references/tagged-errors.md` and make each boundary decision with `references/result-boundaries.md`. Add the proposed dispositions, user-facing mappings, boundary decisions, and ranked vertical migration slices to the report.

Recommend the best first slice and explain why it is bounded and valuable.

Complete when: every catalogued failure has a disposition, every boundary has a recorded codec decision, and one first slice is recommended.

4. Present the approval gate

Present the report and recommendation. The audit branch's only repository change is the report.

Complete when: the report is presented. An explicit approval of one named vertical slice starts the implementation branch; otherwise the run ends here.

Named-slice branch: implementation

1. Establish the slice and API facts

Read the repository instructions, validation commands, installed better-result API, relevant production code, and tests. Trace the named failure source through its immediate callers to its handling or transport boundary. Reuse an existing adoption report when present; otherwise retain the evidence for the implementation summary.

Complete when: the slice boundary, current behavior, failure variants, callers, side effects, and transport boundaries are accounted for, and the available API is verified.

2. Implement one vertical slice

Follow `references/vertical-slice-migration.md`, using `references/tagged-errors.md` for failure dispositions and `references/result-boundaries.md` for boundary contracts. Keep the implementation within the named propagation path.

Complete when: the path is consistently Result-based from failure source to its chosen handling boundary, with every error variant handled and every control-flow seam accounted for.

3. Verify and report

Run the repository's applicable format, lint, type-check, and relevant test commands. Update an existing adoption report with completed work, deviations, discoveries, and validation evidence; otherwise include them in the user-facing summary. End after recommending the next slice.

Complete when: all applicable checks pass, or every failure is reported with evidence, and the report or summary accurately describes the repository's current state.

Security

  • Security scanPassed, 100 of 100Pattern checks on the skill's files.
  • Model reviewNot runA model review is scheduled for the next rescan.

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
2,002
Repository forks
43
7-day trend score
0

Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/dmmulroy/better-result/adopt-better-result/badge)](https://verified-skill.com/skills/dmmulroy/better-result/adopt-better-result)

Something wrong with this skill or its page? Tell us.