Skip to content

Workflows

Salman Ashraf edited this page Sep 23, 2026 · 4 revisions

Workflows

Workflows chain agents and skills into full mobile engineering loops. Use a workflow when a task needs multiple stages, evidence, or handoff.


How To Use A Workflow

Use the workflow at workflows/<name>.md.

GOAL:
<what you want done>

CONTEXT:
<PRD, issue, files, build command, device info, logs, screenshots>

Installed workflows can also be used as slash commands or copied into your AI tool.


All Workflows

Workflow File Use When Output
feature-ship workflows/feature-ship.md Turn a feature ticket into implementation, review, tests, and PR notes Feature shipping plan/report
release-cycle workflows/release-cycle.md Run feature freeze through release checks and rollout Release readiness sequence
game-level workflows/game-level.md Build/review a Unity or Unreal gameplay level flow Game design, perf, test guidance
crash-to-fix workflows/crash-to-fix.md Go from crash alert to fix, regression test, and deploy Root cause, fix, test, monitor plan
ci-setup workflows/ci-setup.md Generate mobile CI/CD configuration GitHub Actions/Bitrise/Fastlane setup
new-screen workflows/new-screen.md Build a new screen from spec/design UI implementation and review loop
new-project-setup workflows/new-project-setup.md Start a new mobile app with architecture, CI, security baseline Setup checklist and starter plan
app-launch workflows/app-launch.md Prepare a mobile app for store launch Security, performance, listing, release checks
perf-sprint workflows/perf-sprint.md Run a focused performance improvement cycle Baseline, fixes, measurements
mobile-flight-recorder workflows/mobile-flight-recorder.md Preserve project context before switching sessions or tools MOBILE_AGENCY_CONTEXT.md and resume prompt
mobile-memory-workflow workflows/mobile-memory-workflow.md Preserve and resume project context Mobile Memory checkpoint
issue-to-agent workflows/issue-to-agent.md Convert a GitHub issue into the right agent/skill execution plan ISSUE EXECUTION PLAN
device-proof-report workflows/device-proof-report.md Turn Mobile MCP evidence into durable proof DEVICE_QA_REPORT.md
appforge-workflow workflows/appforge-workflow.md Go from app idea to PRD, design, tasks, QA, Play Store prep Product and implementation artifacts
mobile-mcp-qa workflows/mobile-mcp-qa.md Verify a mobile flow on a device, emulator, or simulator MOBILE_MCP_QA_REPORT.md
mobile-harness workflows/mobile-harness.md Guide beginners or execute approved tasks through build, verification, device proof, and memory MOBILE_HARNESS_REPORT.md

Device Proof Report

Use device-proof-report after /mobile-mcp-qa when you need evidence a teammate or reviewer can inspect later.

It records:

  • build installed and app launched
  • device, OS, orientation, app id, build, commit
  • actions performed
  • screenshots and what each screenshot proves
  • assertions with PASS/FAIL/BLOCKED results
  • crashes/logs
  • accessibility and performance notes
  • issues with repro steps and suggested fixes

Use it for PR reviews, release gates, bug fixes, screenshot validation, and Mobile Harness task completion.


Mobile Flight Recorder

Use mobile-flight-recorder at the end of a session when decisions, code changes, bugs, device setup, or next actions must survive across AI tools.

It creates or updates MOBILE_AGENCY_CONTEXT.md, a stable project flight recorder that records:

  • app goal, platform, stack, repo, and branch
  • current architecture and implemented features
  • current task, pending tasks, known bugs, and risks
  • test/build commands that actually ran
  • design source and Mobile MCP/device setup
  • agent findings, decisions, changed files, evidence, and links
  • exactly one next recommended action

Run it with:

Run Mobile Flight Recorder. Update MOBILE_AGENCY_CONTEXT.md and MOBILE_MEMORY.md if useful. Capture decisions, files changed, tests run, known bugs, device setup, links, and exactly one Next Recommended Action. Do not include secrets.

Resume any AI tool with:

Read MOBILE_AGENCY_CONTEXT.md and MOBILE_MEMORY.md if present. Continue from Next Recommended Action. Inspect changed files before editing and update the flight recorder before stopping.

Flight Recorder integrates with MOBILE_MEMORY.md, MRECALL.md, and .mobile-ai-agents/memory/ when present. Use MOBILE_AGENCY_CONTEXT.md for the broader project state and MOBILE_MEMORY.md for compact session continuity.


Performance Loop

Use the Performance Loop when the issue is performance-related and needs measurable proof. It covers:

  • startup and time to interactive
  • ANRs and frozen UI
  • memory leaks and OOM risk
  • frame drops and jank
  • battery drain
  • network overuse
  • APK/AAB/IPA or bundle size
  • slow screens
  • regression monitoring

Route the work like this:

Symptom Start With Evidence
Startup delay @PERF, /perf-audit Cold/hot start times, trace
ANR @FREEZE, /anr-investigation Play Console cluster, traces.txt
Memory leak @RETAINER, /memory-leak-investigation LeakCanary trace, heap growth
Jank @PERF, platform reviewer Frame times, profiler trace
React Native re-render performance-optimizer, /rn-performance React Profiler, Flipper
Unity frame spike @FORGE, /game-perf Unity Profiler, frame debugger

The expected handoff is PERFORMANCE_AUDIT_REPORT.md with baseline metrics, findings, root cause, changes made, before/after verification, trace links, and regression monitoring.


Issue To Agent

Use issue-to-agent when a GitHub issue needs to become executable work. It classifies the issue type, chooses the right agent/skills, names likely files, creates an implementation plan, and defines verification and done criteria.

Supported issue types include:

  • new feature
  • crash
  • ANR
  • memory leak
  • UI bug
  • security issue
  • release task
  • store listing
  • CI/CD task
  • documentation

Home · Agents · Skills and Prompts

Clone this wiki locally