# Feature Flags

> >-

- **Type:** Skill
- **Install:** `agentstack add skill-alisinadevelo-md-files-feature-flags`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [AlisinaDevelo](https://agentstack.voostack.com/s/alisinadevelo)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [AlisinaDevelo](https://github.com/AlisinaDevelo)
- **Source:** https://github.com/AlisinaDevelo/md-files/tree/main/plugins/forge/skills/feature-flags

## Install

```sh
agentstack add skill-alisinadevelo-md-files-feature-flags
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Feature Flags

A feature flag decouples *deploy* from *release*: ship the code dark, turn it on when
ready, turn it off instantly if it misbehaves. Powerful — and a debt magnet if you don't
retire them. The single most important rule: **every flag needs a removal plan from the
day it's created.**

## Flag types (don't conflate them)

- **Release flag** — temporary, gates an in-progress feature. Removed once fully rolled out.
- **Ops / kill switch** — long-lived, lets you disable an expensive or risky subsystem under
  load or incident. Kept on purpose.
- **Experiment flag** — drives an A/B test; removed when the experiment concludes.
- **Permission / entitlement** — long-lived, gates by plan/tenant. This is config, not a
  feature flag — treat it as such.

Mixing these is how you end up with 200 flags nobody understands. Name and document which
kind each one is.

## Safe rollout

- **Default off.** A new flag's default is the old behavior. Shipping the code does nothing
  until you deliberately turn it on.
- **Ramp progressively:** internal → 1% → canary cohort → 10% → 50% → 100%, watching the
  error rate and key metrics at each step. Don't jump to 100%.
- **Make it killable instantly** — flipping off must not require a deploy, and off must be a
  safe state at any ramp percentage.
- **Both paths must work at every step.** While a flag is live, old and new code paths both
  run in production. Test both; don't let the off-path rot.

## The cleanup discipline (where teams fail)

- A release flag is **temporary**. When it's at 100% and stable, delete the flag *and the
  dead branch* — the code, the config, the tests for the old path. A flag left in forever is
  a permanent `if` nobody dares remove.
- Track flag age. A release flag older than its feature is a smell. Put an expiry/owner on
  it; review stale flags regularly.
- Avoid **flag dependencies** (flag B only sane when flag A is on) — they multiply states
  combinatorially and make reasoning impossible.

## Gotchas

- **State explosion:** N flags = up to 2^N code paths. Keep the live set small.
- **Consistency:** a user flipping mid-session between on/off paths can hit inconsistent
  state — evaluate the flag once per request/session, not per call.
- **Don't gate data migrations on a release flag** the way you'd gate UI — a half-migrated
  schema behind a flag is a different, riskier problem (see `safe-database-migrations`).

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [AlisinaDevelo](https://github.com/AlisinaDevelo)
- **Source:** [AlisinaDevelo/md-files](https://github.com/AlisinaDevelo/md-files)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-alisinadevelo-md-files-feature-flags
- Seller: https://agentstack.voostack.com/s/alisinadevelo
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.
