agentleFS
LedgerHQLedgerHQ/ledger-live/.agents/skills/remove-feature-flag/SKILL.md

remove-feature-flag

Skill627 starsChanged 8 days agoMITView on GitHub

Remove a Ledger Wallet feature flag while preserving its resolved production behavior, then validate, create a draft PR, and watch CI to green. Use when the user runs /remove-feature-flag, and only then; it may stash local work, push a new branch, create or update Jira work, and publish a PR.

7 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use classify-changelog-pr, cross-platform-cloud-verification and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • black12-agSkill

    classify-changelog-pr

    Reference guidance for classifying whether an unmarked PR should appear in the changelog and under which category. Used inline by the changelog-draft skill — not dispatched as a separate agent.

    Also in 4 projects that use this

  • black12-agSkill

    cross-platform-cloud-verification

    Orchestrates cost-conscious cloud verification across available operating-system and architecture runners after cheaper checks pass. Use whenever a code change, bug fix, build, packaging flow, native dependency, UI behavior, filesystem behavior, or test has material cross-platform implications, even if the user only asks to verify or test the change. Consult domain verification skills for what to test; use this skill to choose and access the smallest relevant set of platforms.

    Also in 4 projects that use this

  • +10 more from black12-ag

The file

Show its rules and text

We split this skill into the rules it gives an agent

17 rules. 0 are common, 10 are distinctive, and 7 are very specific to this project.

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 10

    Distinctive

    This author's own choices. Worth a look.

  • 7

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Run the complete removal workflow.

    line 13
  2. Very specific:

    Only FEATURE_ID is required.

    line 27
  3. Distinctive:

    If the lookup is unavailable, missing, malformed, or unauthorized, ask the user for the production value.

    line 48
  4. Distinctive:

    If Firebase reports conditional values, or the JSON describes A/B variants, show them and ask which behavior to preserve.

    line 50
  5. Very specific:

    Create it directly from the freshly fetched origin/develop.

    line 69
  6. Distinctive:

    When no ticket is supplied and Jira tools are available:

    line 77
  7. Distinctive:

    Search the owning team's backlog by feature id and Firebase key.

    line 79
  8. Very specific:

    If none match, create an unassigned Task in project LIVE for the owning team.

    line 81
  9. Very specific:

    If Jira is unavailable, continue without a ticket and add this to the PR description: No Jira linked: Jira access was unavailable during PR …

    line 84
  10. Distinctive:

    Search tracked source across the whole repository for the feature id and Firebase key.

    line 89
  11. Distinctive:

    Delete the flag definition and barrel export.

    line 93
  12. Very specific:

    Remove legacy types-live, defaults, devtools, analytics, E2E, mocks, and test references when present.

    line 94
  13. Distinctive:

    Update focused tests and snapshots without weakening coverage.

    line 97
  14. Distinctive:

    Keep the change isolated from unrelated cleanup.

    line 98
  15. Very specific:

    Use validate-changes for the affected packages and files, including @shared/feature-flags checks when its registry changes.

    line 105
  16. Very specific:

    Use fix-ci for related failures, with at most three diagnose/fix/push cycles.

    line 126
  17. Distinctive:

    When required draft checks pass, mark the PR Ready for review.

    line 130

“Common” is measured: kept by at least 6 in 10 of the 7 edited versions of this file we could read. “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

Remove Feature Flag

Run the complete removal workflow. Never write to Firebase.

Invocation

/remove-feature-flag <FEATURE_ID> [FLAG_VALUE_JSON] [TICKET_URL]

Optional modifiers:

  • proposal only — analyze and propose without changing files
  • no pr — implement and validate without publishing
  • no watch — create the PR without waiting for CI

Only FEATURE_ID is required. Treat invoking this command as authorization for the complete workflow unless a modifier limits it. Accept optional arguments in any trailing order: recognize the JSON object, an http(s) ticket URL, and the exact modifier phrases independently.

For proposal only, stop after read-only analysis. Do not create Jira work, stash files, create a branch, commit, push, or publish a PR. Report the resolved behavior, owner, affected scope, proposed edits, planned validation, and skipped Jira/PR/CI actions.

1. Resolve the production behavior

Derive the Firebase key with the repository's canonical snake-case mapping.

  1. Parse and use FLAG_VALUE_JSON when supplied. Validate it against the flag definition and consumers. If it is invalid, lacks a boolean enabled, or omits params that code consumes, ask for a corrected production value.
  2. Otherwise run:
    node .agents/skills/remove-feature-flag/scripts/read-production-flag.mjs "$FEATURE_ID"
  3. If the lookup is unavailable, missing, malformed, or unauthorized, ask the user for the production value. Never fall back to development.
  4. If Firebase reports conditional values, or the JSON describes A/B variants, show them and ask which behavior to preserve. Do not choose a client-specific variant automatically.

Preserve the resolved production behavior whether enabled is true or false, including the selected production params.

2. Protect local work and create the branch

Record the initial branch/ref and worktree state. If dirty, stash staged, unstaged, and untracked files with a unique descriptive message; do not ask.

Fetch the latest remote base and always create a new branch, named per git-conventions (chore, never the legacy support):

chore/feature-flags-remove-<feature-id>-flag-<YYYYMMDD-HHmm>

Create it directly from the freshly fetched origin/develop. If the name collides, append a numeric suffix. Never reuse an existing branch or PR.

3. Resolve Jira context

Identify the owning team from flags/team-* and CODEOWNERS; use the codeownership skill when a team directory is involved.

When no ticket is supplied and Jira tools are available:

  1. Search the owning team's backlog by feature id and Firebase key.
  2. Reuse one clearly relevant removal ticket; ask if several match.
  3. If none match, create an unassigned Task in project LIVE for the owning team.

If Jira is unavailable, continue without a ticket and add this to the PR description: No Jira linked: Jira access was unavailable during PR creation.

4. Remove the flag

Search tracked source across the whole repository for the feature id and Firebase key; ignore generated and build artifacts. Load and use all applicable repository skills for the files involved.

  • Delete the flag definition and barrel export.
  • Remove legacy types-live, defaults, devtools, analytics, E2E, mocks, and test references when present.
  • Inline only the selected production behavior; remove unreachable branches.
  • Update focused tests and snapshots without weakening coverage.
  • Keep the change isolated from unrelated cleanup.

Firebase remains read-only. Add a post-release PR note to delete the remote key manually.

5. Validate and publish

Use validate-changes for the affected packages and files, including @shared/feature-flags checks when its registry changes. When a package has no nx target for a step, read its README and package.json and run the equivalent scoped command directly.

Unless no pr is set, invoke create-pr end-to-end as the single source of truth for Git conventions, changesets, commits, push, draft PR content, browser opening, and the generated (not posted) Slack message. Pass it the resolved behavior, Firebase key, Jira result, validation evidence, affected flows, and the manual Firebase cleanup note.

For no pr, follow git-conventions, create a local commit after validation, and do not push. Preserve that commit on the removal branch before restoring the initial workspace.

6. Watch CI

Skip this section for no pr or no watch.

  1. Watch required checks on the draft PR. Accept skipped or neutral only for optional checks.
  2. Use fix-ci for related failures, with at most three diagnose/fix/push cycles.
  3. Re-run an apparently unrelated failure once. If it fails again, report it and stop without changing unrelated code.
  4. When required draft checks pass, mark the PR Ready for review.
  5. Continue watching checks triggered by that transition under the same rules.
  6. Finish only when all required checks are green, or when a guardrail stops the workflow.

7. Restore the workspace

On success or early stop, preserve any removal work on its branch or a recovery stash, then return to the initial branch/ref and restore the saved stash so the workspace matches its starting state. Never discard work. If restoration conflicts, stop and report the conflict without resolving unrelated user changes.

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.