Should Flag Change Skill

An advisory, read-only Agent Skill that decides whether a given code change should be released behind a LaunchDarkly feature flag.

Overview

This skill teaches agents how to:

Read a git diff (or in-progress change) and understand what it actually does
Explore the surrounding code to judge blast radius and behavior impact
Weigh the change against a decision framework (favoring flags for user-facing and risky changes)
Emit a single structured verdict via the recommend-flag tool

It is advisory and read-only: it never creates, toggles, or modifies flags.

Installation (Local)

For now, install by placing this skill directory where your agent client loads skills.

Generic: copy skills/feature-flags/should-flag-change/ into your client's skills path

Prerequisites

None. This skill does not require the LaunchDarkly MCP server. It reads code with standard file tools and returns a structured recommendation via the recommend-flag tool.

Usage

Ad hoc, while working on a change:

Should this change be behind a feature flag?

In CI on a pull request, the skill is fed the PR's git diff and reads the repository to assess it, then emits a recommend-flag verdict ({ recommend, verdict, confidence, reasons }) that a check can parse.

Output

The skill ends by calling the recommend-flag tool:

{
  "recommend": true,
  "verdict": "suggested",
  "confidence": "medium",
  "reasons": [
    "New public endpoint added in src/routes/export.ts — user-facing path with no existing gate"
  ]
}

recommend is the boolean a CI check keys on. verdict is the finer signal: suggested (a new flag is warranted), reuse-existing (gate the change behind an existing flag that already covers this feature — set reuse_flag_key, don't create a new one), already-flagged (protected by an existing / ancestor flag — covered, no action), or not-suited (genuinely nothing to flag). Keeping these outcomes distinct is what lets a dashboard track real flag coverage.

Structure

should-flag-change/
├── SKILL.md
├── marketplace.json
└── README.md

LaunchDarkly Flag Create: Create the flag once you've decided to
LaunchDarkly Flag Cleanup: Remove flags when they're no longer needed
LaunchDarkly Docs

License

Apache-2.0

Third-party checks

No alerts1mo3 checks · Risk SAFE

Gen Agent Trust Hub1mo
This skill is a secure, read-only advisor from LaunchDarkly that assists developers in evaluating whether code changes should be released behind feature flags. It analyzes git diffs and repository content without the ability to modify code, flags, or system state, and does not require external network access.
Socket1mo
No alerts
Snyk1mo
Risk: LOW · No issues

Capability

metadata: { "author": "launchdarkly", "version": "0.4.0-experimental" }

Other metadata

compatibility: Advisory and read-only. Works on all platforms. Does NOT require the LaunchDarkly MCP server. Reads code with standard file tools (Read/Grep/Glob) and returns a structured verdict via the `recommend-flag` tool.

/should-flag-change

README.md