Audit Better Result Dependents
Testing & Qualityaudit-better-result-dependentsAudit Better Result Dependents is an Agent Skill for testing & quality published in the dmmulroy/better-result repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.
Audit better-result changes or PRs against known Prisma and Better T Stack downstream dependents. Use when working on better-result API/type/runtime changes and the user asks whether a change breaks Prisma dependents, Better T Stack dependents, npm dependents, or PR compatibility.
How to use this skill
Bring this guide into your coding agent with a prompt tailored to the tool you use.
- Open your project in Codex.
- Copy and paste the prompt below.
- Review files and risks before approving any action.
Inspect the Agent Skill at https://github.com/dmmulroy/better-result/blob/HEAD/.agents/skills/audit-better-result-dependents/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/audit-better-result-dependents, but wait for my approval before writing files or running scripts.
Copying this prompt does not install or run the skill. Review third-party files before use.
Audit better-result dependents
Quick start
From the better-result repo, run the audit script with identifiers touched by the change:
node .agents/skills/audit-better-result-dependents/scripts/audit-dependents.mjs isTaggedError 'TaggedError\.is' 'toJSON'
If no patterns are provided, the script searches common better-result usage patterns.
Scope
Always check these npm dependents:
@prisma/compute-sdk@prisma/streams-server@prisma/streams-localcreate-better-t-stack@better-t-stack/template-generator
The script uses npm registry metadata to resolve latest versions and commits. For public GitHub repos it checks out the published gitHead; if a repo is private/unavailable, it falls back to npm pack and audits the published source/package.
Workflow
-
Identify risk surface
- List changed public exports, guards, types, class methods, overloads, and runtime semantics.
- Convert them into grep patterns, escaping regex metacharacters when needed.
-
Run local validation first
bun run fmtbun run checkbun run testbun run lint
-
Audit dependents
- Run the script with risk patterns.
- Inspect actual matching files with
read; do not rely only on counts. - Distinguish generic guards (e.g.
isTaggedError) from class-specific guards (e.g.FooError.is).
-
Assess breakage
- Runtime breaking: dependent passes values that no longer satisfy a changed guard/shape.
- Type breaking: dependent references narrowed types, overloads, exported names, or structural constraints that changed.
- Safe: dependent only creates
TaggedError(...)instances or uses class-specific.is()unaffected by the change.
-
Report clearly
- Include package name, version, source audited (GitHub commit or npm package), relevant matches, and risk conclusion.
- If opening a PR, include the dependent audit summary in the PR body.
Notes
@prisma/compute-sdkmay point at a private/404 GitHub repo; audit the npm package source when cloning fails.@prisma/streams-serverand@prisma/streams-localshare theprisma/streamsrepo and usually the samegitHead.create-better-t-stackand@better-t-stack/template-generatorshare theAmanVarshney01/create-better-t-stackrepo and usually the samegitHead.- Never modify cloned dependent repositories during audit; use temp directories only.