feature-flag-onboarding
Orchestrate first-time Harness FME onboarding for one application. Confirm an already-enabled FME account/project, then pick one verified non-production environment; if none exists, hand off to an admin, not creating one. Detect the codebase's language and framework, then delegate code integration to instrument-feature-flag — never duplicating its wrapper design or creating the same flag twice. Verify two distinct named-treatment outcomes live through separately approved targeting changes when needed, not just default/fallback responses. Resumes from live state and a non-secret progress summary instead of a checkpoint file; never recreates a flag from uncertain absence or reuses a stale approval. Use when asked to onboard FME for an app, set up feature flags end-to-end, or prove a flag works live. Not for flag creation alone (create-feature-flag), code gating alone (instrument-feature-flag), targeting alone (update-flag-targeting), or metric/experiment setup.
Harness · Static checks on file
Installation
Read the source and dependencies before running a command. Copying does not install or execute anything.
npx skills add harness/harness-skills --skill feature-flag-onboardingAbout this skill
name: feature-flag-onboarding description: >- Orchestrate first-time Harness FME onboarding for one application. Confirm an already-enabled FME account/project, then pick one verified non-production environment; if none exists, hand off to an admin, not creating one. Detect the codebase's language and framework, then delegate code integration to instrument-feature-flag — never duplicating its wrapper design or creating the same flag twice. Verify two distinct named-treatment outcomes live through separately approved targeting changes when needed, not just default/fallback responses. Resumes from live state and a non-secret progress summary instead of a checkpoint file; never recreates a flag from uncertain absence or reuses a stale approval. Use when asked to onboard FME for an app, set up feature flags end-to-end, or prove a flag works live. Not for flag creation alone (create-feature-flag), code gating alone (instrument-feature-flag), targeting alone (update-flag-targeting), or metric/experiment setup. metadata: author: Harness version: 1.0.1 mcp-server: harness-mcp license: Apache-2.0 compatibility: Requires the Harness MCP server or the Harness CLI, plus local application code access and (for live proof) runtime/build access; guidance-only or partial if application/runtime access is unavailable
Feature Flag Onboarding
Orchestrate a new application's first end-to-end FME setup: confirm scope, pick a safe environment, detect the codebase, delegate code integration, then prove two distinct treatments actually run live. This skill sequences and verifies; it does not re-implement flag creation, code wiring, or targeting changes — those live in /create-feature-flag, /instrument-feature-flag, and /update-flag-targeting.
Tools
Works through the Harness MCP server or the Harness CLI; names are from tool-map.md. All writes are delegated to the linked skills — this skill only reads.
| Operation | MCP | CLI |
|---|---|---|
| List environments | harness_list · fme_environment · compact: false | harness list fme_environment --json |
| List flags | harness_list · fme_feature_flag · size: 50 · compact: false | harness list feature_flag --json --limit 50 |
| Get flag | harness_get · fme_feature_flag · params.feature_flag_name | harness get feature_flag <name> --json |
| Get definition | harness_get · fme_feature_flag_definition · params: { feature_flag_name, environment_id } | harness get feature_flag:definition <flag> --env <env-id> --json |
Instructions
Phase 1: Scope and account check
Follow scope-establishment.md, separating confirmed account/org/project from the application repository. Establish the supported CLI or MCP transport per mcp-setup.md before API calls; CLI users need not configure MCP. This workflow starts with an existing FME-enabled account/project. Verify the needed read capability in that scope. A 401/403, missing tool or empty inventory does not by itself prove missing entitlement: diagnose the actual error without changing scope/transport to bypass it. Missing account/project/enablement prerequisites go to the authorized user/admin, not automatic provisioning.
Phase 2: Resume check (no checkpoint file)
Use any non-secret progress summary to identify scope/app/flag, then re-read current state; a summary is context, not proof or approval. Ask about ambiguous identities before choosing a candidate. Prefer Get flag/definition for confirmed IDs; use paginated discovery only when needed. A 404 requires scope/identifier/access checks, not automatic creation. Do not infer application integration or live success from a flag merely existing.
Track transport, environment, SDK/configuration, flag/definition, app integration/tests, outcomes A/B and final state as verified / incomplete / unknown, with evidence. During Phase 4/6 inspect the actual app revision, SDK initialization/wrapper and tests as well as remote state. Resume the first incomplete phase without repeating completed mutations; revalidate proof if revision, environment, identity, treatments or targeting changed. Never reuse stale approval. Guidance-only/missing runtime access remains partial.
Phase 3: Select one non-production environment
Reuse a complete List environments inventory and verify the explicitly selected environment's ID and isProduction: false; do not ask again when selection is already clear. Otherwise recommend an isolated/test environment and confirm it. Follow pagination before concluding none exist. With no non-production environment, stop for an admin handoff—never create one or silently use production.
Phase 4: Detect the codebase, then clarify
Make this sequence visible: detect → clarify → delegate the wrapper plan. Do a light detection pass per language-guidance.md (language/framework/runtime, monorepo layout, existing FME/OpenFeature/Classic-FF usage) — enough to pick the target app and know whether /instrument-feature-flag will find an existing wrapper. If the installed flag system can't be identified (Classic FF only, or an unverified OpenFeature provider), stop and resolve that before delegating. Summarize what was detected, then ask only for unresolved choices: target app/service, purpose and candidate gating point, evaluation identity/traffic type, intended treatment outcomes/fallback and customer extension needs. Existing definitions constrain valid treatment names; for a new flag, propose names such as on/off only for explicit confirmation, never silently assume them. Do not design the wrapper, SDK recipe, or treatment mapping here — that belongs entirely to /instrument-feature-flag in Phase 6. Carry every confi