Index of every page on this site: llms.txt
All skills
launchdarkly avatar

/should-flag-change

@f0ee110 official
Run it yourself

Decide whether a committed code change should ship behind a LaunchDarkly feature flag. Use when a developer asks whether a change should be behind a flag, when reviewing a diff or pull request, or when running in CI on a PR. Runs a fixed decision tree over the change and the code around it, with checked evidence, and emits a structured advisory recommendation. Run it on every change, including test-only, docs, dependency and already-flagged ones: those finish in a minute or two, and its verdict, not your own read of the diff, is the decision. Read-only: it never creates or modifies flags.

Use this Skill: https://skilld.dev/gh/launchdarkly/ai-tooling/should-flag-change

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

referencesCONSUMERS.md

≈1.2k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Consumer search

One question about this change couldn't be settled from the evidence: whether it changes what an EXISTING interface returns in a way a SEPARATE client reads. The interface isn't shown to be published, and the evidence shows no client reading what changed; that client is often in another repository. Find it, or show there is none. Not a judgment that a client "probably" depends on it: a reader you can show, line by line, in the code.

THE CHANGE

  • Repository: {CHECKOUT}. Read files at the change's commit only: git -C {CHECKOUT} show {HEAD}:<path>; search with git -C {CHECKOUT} grep -n <pattern> {HEAD} -- <paths>. The working tree may hold other work: don't read it.
  • The change's commit: {HEAD}. The version on main it is compared with: {BASE}. See the change with git -C {CHECKOUT} diff {BASE}...{HEAD}.
  • The evidence, including the PR text: {RUN}/evidence.txt (its CHANGED FILES section gives each file's role; a PUBLISHED API LOOKUP section, when present, says which changed operations are in the published API).

CLIENT REPOSITORIES Checkouts of other repositories whose code calls this one, each as it was on its main branch when the change was compared (a client written after the change can't show what relied on the old contract): {CLIENTS}

WHAT IS A CONTRACT CHANGE A behavior-carrying hunk that changes a status code, error shape or error code, field presence or type, pagination, ordering, or a limit, in what an interface that exists at {BASE} returns (an HTTP or GraphQL response, a tool or RPC result, a webhook payload, stored data another program reads).

  • A request the change makes differently (what it sends) is not one.
  • An interface the diff adds has no contract yet.
  • A response that failed at {BASE} and now succeeds is a repair, not one. Except a guard: a change that removes or loosens a check that refuses an action on purpose -- an approval requirement, a permission or role check, a plan, entitlement or policy limit -- so requests it was meant to refuse at {BASE} now go through IS one. A guard that refused requests it was meant to allow (the PR or a test the diff adds shows the refusal was the bug) is repaired, not removed. A validation error, a lookup that wrongly found nothing, or a crash is not a guard.
  • A request that still fails, where only how the failure is reported changes (a server error, 5xx, that now returns a client error, 4xx, or a clearer error message or body), is a repair too. It IS one when the failure moves the other way (a client error that now returns a server error), or when a separate reader at {BASE} branches on the old status or error code.

WHAT IS A SEPARATE READER Code in a separate client -- a frontend, SDK, CLI, another service, or a client repository -- that reads the specific thing that changed: the field it accesses, the status code or error code it branches on, the order or page it relies on. Code in the same program as the server (its own handlers, its own tests) is not a separate reader. Code that only calls the interface without reading what changed is not a reader of the change.

WHAT TO REPORT Report each contract change. If you find a separate reader of it, give a chain of real lines, quoted exactly:

  • at least one CHANGED line: a line the diff adds or modifies, in shipping code (role "changed", repo "self");
  • the READER line: the client line that reads the changed field, status or code (role "reader", repo "self" or the client repository's name above);
  • "token": the exact identifier, field name, status code or error code the reader reads. It must appear in the reader line and in the diff. Every link is checked mechanically: the changed line must be in the diff, the reader line must exist at its commit outside tests, stories and docs, be in a client repository or in a different language from the changed line, and contain the token, and the token must appear in the diff. A link that doesn't match drops the reader. Report a contract change with an empty chain when you searched and found no reader: that is a normal, useful answer, and it settles the question as no. If the change alters no contract at all, return an empty list.

Read-only: never modify any checkout, use the network, or read other folders. Never print a token or key.

Write ONLY this JSON to {RUN}/consumers.json:

{"contract_changes": [{"interface": "the existing interface (route, query, tool, payload)", "what_changed": "one sentence: the status, field, error, order or limit that changed", "token": "the identifier a reader reads, or empty when no reader was found", "chain": [{"repo": "self | <client name>", "path": "path/at/that/commit", "line": 123, "text": "the exact line", "role": "changed | reader | calls"}], "searched": "where you looked for readers"}]}

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

1 alerttoday3 checks · Risk MEDIUM
  • Gen Agent Trust Hubtoday

    This skill utilizes a bundled executable that contains hidden Unicode characters, which could be used for malicious steganography, and it analyzes untrusted data from repositories and pull requests, making it susceptible to indirect prompt injection.

  • Sockettoday

    1 alert: gptSecurity

  • Snyktoday

    Risk: CRITICAL · 2 issues

Signed by skilld at f0ee110. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated 2 days ago
metadata
{
  "author": "launchdarkly",
  "version": "0.5.3",
  "graph": "07355490e985"
}
Other metadata
compatibility
Advisory and read-only. Needs a git checkout with the change committed (shallow is fine; a --filter partial clone is filled in from its remote), a shell, git and Python 3.10+ (no packages). Reads LaunchDarkly flag state through your LaunchDarkly tools when available, and works without them. When it can't run (no shell, no checkout), it decides by judgment instead, from references/NO_CHECKOUT.md. Returns a structured verdict as a `recommend-flag` block.

README badge

README badge for launchdarkly/ai-tooling/should-flag-change