Migrate @commitlint/cz-commitlint to @inquirer/prompts to remove inquirer v12 peer cap

[taterhead247]

Summary

@commitlint/cz-commitlint currently constrains inquirer to ^9 || ^10 || ^11 || ^12, which blocks downstream repos from upgrading their workspace inquirer reference to v13+ while using this adapter.

This issue documents:

• what was verified,
• why downstream installs are blocked,
• and the planned migration path from legacy inquirer APIs to @inquirer/prompts for @commitlint/cz-commitlint.

Verified Findings

1) Current peer dependency cap in @commitlint/cz-commitlint

In @commitlint/cz-commitlint/package.json, peer deps currently include:

• inquirer": "^9.0.0 || ^10.0.0 || ^11.0.0 || ^12.0.0"

This is the direct compatibility cap for consumers using newer inquirer majors.

2) inquirerIns in prompter() is not truly “unused” today

Code comment says:

@param inquirerIns instance passed by commitizen, unused

But code currently passes it through and calls .prompt(...):

• src/index.ts passes inquirerIns to process(...)
• src/Process.ts does await inquirer.prompt(questions)

So this comment is stale/misleading. More accurate wording: only the prompt function shape is used, not the full inquirer API surface.

Why migrate to @inquirer/prompts

The legacy monolithic inquirer package is still maintained but is not the actively developed API surface. @inquirer/prompts is the modern prompt API and supports progressive adoption patterns.

Migrating @commitlint/cz-commitlint to @inquirer/prompts would remove the tight coupling to old inquirer question typings (DistinctQuestion, Answers, etc.) and avoid the peer cap mismatch for consumers.

Proposed Migration Plan (PR scope)

Goal

Migrate only @commitlint/cz-commitlint prompt execution path from inquirer.prompt(questions[]) to @inquirer/prompts while preserving current UX and output behavior.

Key implementation strategy

1. Introduce a package-local prompt model

• Replace direct inquirer runtime/type coupling in src/Question.ts, src/Process.ts, section builders, and tests.
• Keep existing validation/filter/transformation semantics.

2. Replace prompt engine

• Map old prompt types to modern prompts:
  • input -> input
  • list -> select
  • checkbox -> checkbox
  • confirm -> confirm
• Preserve conditional when behavior and sequential flow currently provided by array-based .prompt(...).

3. Preserve separator/choice UX

• Current code uses new inquirer.Separator() in enum flows.
• @inquirer/prompts supports Separator in select / checkbox; adopt equivalent usage.

4. Keep Commitizen adapter contract stable

• Exported prompter(...) API should remain compatible for users invoking through commitizen.
• Internally, parameter can be ignored or kept for signature compatibility during transition.

5. Update dependencies/docs

• Remove old inquirer peer constraint from @commitlint/cz-commitlint once no longer needed.
• Add/update deps for @inquirer/prompts.
• Update adapter README install guidance that currently recommends legacy inquirer explicitly.

File areas expected to change

• @commitlint/cz-commitlint/package.json
• @commitlint/cz-commitlint/src/index.ts
• @commitlint/cz-commitlint/src/Process.ts
• @commitlint/cz-commitlint/src/Question.ts
• @commitlint/cz-commitlint/src/SectionHeader.ts
• @commitlint/cz-commitlint/src/SectionBody.ts
• @commitlint/cz-commitlint/src/SectionFooter.ts
• @commitlint/cz-commitlint/src/*.test.ts
• @commitlint/cz-commitlint/README.md

Testing Plan

1. Unit parity

• Ensure existing tests around:
  • required/optional fields,
  • max/min validations,
  • case/full-stop filters,
  • conditional footer/body prompts,
  • combined commit message output
    continue to pass or are updated with equivalent expectations.

2. Integration smoke

• run commitizen flow against adapter in this repo and validate resulting commit messages remain unchanged for canonical cases.

3. Downstream verification (optional but useful)

• locally test a consumer workspace with inquirer v13+ and @commitlint/cz-commitlint from branch/release candidate.

Risk Areas

• Small UX differences between old and new prompt renderers (especially transformer rendering and list behavior).
• Test harness changes required because current tests mock inquirer question object execution.
• Potential need to keep backward-compatible adapter entry signature while decoupling runtime from inquirer.

Non-goals

• Full migration of other commitlint packages (@commitlint/prompt, @commitlint/prompt-cli, etc.) in the same change.
• Broad prompt feature redesign.

Expected Outcome

After this migration, downstream repositories should no longer be blocked by @commitlint/cz-commitlint’s legacy peer range when upgrading their inquirer-related toolchain.

---

If maintainers are open to this direction, I can follow up with a PR scoped exactly to @commitlint/cz-commitlint and keep the change narrowly focused on parity + compatibility.

[taterhead247]

I am so sorry. I asked AI to write up a plan to submit a PR to migrate you guys to @inquirer/prompt so I could remove the inquirer:12 devDependency form one of my repos. Instead of creating the issue on my fork, it added it here. So helpful. And it looks like I can't delete the Issue.

[escapedcat]

Reply in cooperation with Claude:

Thanks for the thorough writeup and for digging into this! 🙏

One thing before anyone starts on a big migration: I don't think we need to move @commitlint/cz-commitlint off the classic inquirer API to lift the peer cap. inquirer.prompt([...]) and inquirer.Separator — which is all this adapter actually uses — still exist in inquirer v13 and v14. The monolithic package is maintenance-mode but the classic interface isn't gone, so the full port to @inquirer/prompts (rewriting Process.ts / Question.ts / the section builders + tests) isn't necessary here.

What does break on v13+ is smaller and specific: the list prompt type was a backward-compat alias through v12 but was removed in the v13 line — it's select now. We emit type: "list" for single-select enums, so that needs to become "select". There's also a little type fallout (ChoiceCollection / InputQuestionOptions are no longer exported, and DistinctQuestion is stricter about name / message).

I tried this locally against inquirer 14.0.2 (v13 is already superseded, so let's target 14): widening the peer range + list→select + the small type fixups is about 24 lines across Question.ts and its test — tsc -b is clean and all 122 cz-commitlint tests pass.

So if you'd like to open a PR, the focused scope would be:

• peerDependencies.inquirer: add || ^13.0.0 || ^14.0.0
• Question.ts: type: "list" → "select", drop the removed ChoiceCollection import, and adjust the question construction for the stricter
  types
• engines.node: bump the floor to >=22.13.0 (inquirer 13/14 require it)

No need to touch the prompt flow or section builders. Happy to review once it's up!