Flow-Next Setup Workflow

Flow-Next Setup Workflow

Follow these steps in order. This workflow is idempotent - safe to re-run.

Step 0: Resolve plugin path

The plugin root is the parent of this skill's directory. From this SKILL.md location, go up to find scripts/ and .claude-plugin/.

Example: if this file is at ~/.claude/plugins/cache/.../flow-next/0.3.12/skills/flow-next-setup/workflow.md, then plugin root is ~/.claude/plugins/cache/.../flow-next/0.3.12/.

Store this as PLUGIN_ROOT for use in later steps.

Step 1: Initialize .flow/

Use flowctl init (idempotent - safe to re-run, handles upgrades):

"${PLUGIN_ROOT}/scripts/flowctl" init --json

This creates/upgrades:

• .flow/ directory structure (epics/, specs/, tasks/, memory/)
• meta.json with schema version
• config.json with defaults (merges new keys on upgrade)

Step 2: Check existing setup

Read .flow/meta.json and check for setup_version field.

Also read plugin version from ${PLUGIN_ROOT}/.claude-plugin/plugin.json (Claude Code) or ${PLUGIN_ROOT}/.factory-plugin/plugin.json (Factory Droid) - check whichever exists.

If setup_version exists (already set up):

• If same version: tell user "Already set up with v. Re-run to update docs only? (y/n)"
  • If yes: skip to Step 6 (docs)
  • If no: done
• If older version: tell user "Updating from v to v" and continue

If no setup_version: continue (first-time setup)

Step 3: Create .flow/bin/

mkdir -p .flow/bin

Step 4: Copy files

IMPORTANT: Do NOT read flowctl.py - it's too large. Just copy it.

Copy using Bash cp with absolute paths:

cp "${PLUGIN_ROOT}/scripts/flowctl" .flow/bin/flowctl
cp "${PLUGIN_ROOT}/scripts/flowctl.py" .flow/bin/flowctl.py
chmod +x .flow/bin/flowctl

Then read templates/usage.md and write it to .flow/usage.md.

Step 5: Update meta.json

Read current .flow/meta.json, add/update these fields (preserve all others):

{
  "setup_version": "<PLUGIN_VERSION>",
  "setup_date": "<ISO_DATE>"
}

Step 6: Configuration Questions

6a: Detect current config and tools

Before asking questions, detect available tools and read current config:

# Detect available review backends
HAVE_RP=$(which rp-cli >/dev/null 2>&1 && echo 1 || echo 0)
HAVE_CODEX=$(which codex >/dev/null 2>&1 && echo 1 || echo 0)

# Read current config values if they exist
CURRENT_BACKEND=$("${PLUGIN_ROOT}/scripts/flowctl" config get review.backend --json 2>/dev/null | jq -r '.value // empty')
CURRENT_MEMORY=$("${PLUGIN_ROOT}/scripts/flowctl" config get memory.enabled --json 2>/dev/null | jq -r '.value // empty')
CURRENT_PLANSYNC=$("${PLUGIN_ROOT}/scripts/flowctl" config get planSync.enabled --json 2>/dev/null | jq -r '.value // empty')
CURRENT_CROSSEPIC=$("${PLUGIN_ROOT}/scripts/flowctl" config get planSync.crossEpic --json 2>/dev/null | jq -r '.value // empty')

Store detection results for use in questions. When showing options, indicate current value if set (e.g., "(current)" after the matching option label).

6b: Check docs status

Read the template from templates/claude-md-snippet.md.

For each of CLAUDE.md and AGENTS.md:

1. Check if file exists
2. If exists, check if <!-- BEGIN FLOW-NEXT --> marker exists
3. If marker exists, extract content between markers and compare with template

Determine status for each file:

• missing: file doesn't exist or no flow-next section
• current: section exists and matches template
• outdated: section exists but differs from template

6c: Show current config notice

If ANY config values are already set, print a notice before asking questions:

Current configuration:
- Memory: <enabled|disabled> (change with: flowctl config set memory.enabled <true|false>)
- Plan-Sync: <enabled|disabled> (change with: flowctl config set planSync.enabled <true|false>)
- Plan-Sync cross-epic: <enabled|disabled> (change with: flowctl config set planSync.crossEpic <true|false>)
- Review backend: <codex|rp|none> (change with: flowctl config set review.backend <codex|rp|none>)

Only include lines for config values that are set. If no config is set, skip this notice.

6d: Build questions list

Build the questions array dynamically. Only include questions for config values that are NOT already set.

Available questions (include only if corresponding config is unset):

Memory question (include if CURRENT_MEMORY is empty):

{
  "header": "Memory",
  "question": "Enable memory system? (Auto-captures learnings from NEEDS_WORK reviews)",
  "options": [
    {"label": "Yes (Recommended)", "description": "Auto-capture pitfalls and conventions from review feedback"},
    {"label": "No", "description": "Disable with: flowctl config set memory.enabled false"}
  ],
  "multiSelect": false
}

Plan-Sync question (include if CURRENT_PLANSYNC is empty):

{
  "header": "Plan-Sync",
  "question": "Enable plan-sync? (Updates downstream task specs after implementation drift)",
  "options": [
    {"label": "Yes (Recommended)", "description": "Sync task specs when implementation differs from original plan"},
    {"label": "No", "description": "Disable with: flowctl config set planSync.enabled false"}
  ],
  "multiSelect": false
}

Plan-Sync cross-epic question (include if CURRENT_PLANSYNC is "true" AND CURRENT_CROSSEPIC is empty):

{
  "header": "Cross-Epic",
  "question": "Enable cross-epic plan-sync? (Also checks other open epics for stale references)",
  "options": [
    {"label": "No (Recommended)", "description": "Only sync within current epic. Faster, avoids long Ralph loops."},
    {"label": "Yes", "description": "Also update tasks in other epics that reference changed APIs/patterns."}
  ],
  "multiSelect": false
}

Review question (include if CURRENT_BACKEND is empty):

{
  "header": "Review",
  "question": "Which review backend for Carmack-level reviews?",
  "options": [
    {"label": "Codex CLI", "description": "Cross-platform, uses GPT 5.2 High for reviews. Simple setup, works everywhere. <detected if HAVE_CODEX=1, (not detected) if HAVE_CODEX=0>"},
    {"label": "RepoPrompt", "description": "macOS only. Auto-discovers git diffs + context, reviews scoped to actual changes, ~65% fewer tokens than traditional approaches. <detected if HAVE_RP=1, (not detected) if HAVE_RP=0>"},
    {"label": "None", "description": "Skip reviews, can configure later with --review flag"}
  ],
  "multiSelect": false
}

Docs question (always include):

{
  "header": "Docs",
  "question": "Update project documentation with Flow-Next instructions?",
  "options": [
    {"label": "CLAUDE.md only", "description": "Add flow-next section to CLAUDE.md"},
    {"label": "AGENTS.md only", "description": "Add flow-next section to AGENTS.md"},
    {"label": "Both", "description": "Add flow-next section to both files"},
    {"label": "Skip", "description": "Don't update documentation"}
  ],
  "multiSelect": false
}

Star question (always include):

{
  "header": "Star",
  "question": "Flow-Next is free and open source. Star the repo on GitHub?",
  "options": [
    {"label": "Yes, star it", "description": "Uses gh CLI if available, otherwise shows link"},
    {"label": "No thanks", "description": "Skip starring"}
  ],
  "multiSelect": false
}

Use AskUserQuestion with the built questions array.

Note: If docs are already current, adjust the Docs question description to mention "(already up to date)" or skip that question entirely.

Note: If neither rp-cli nor codex is detected, add note to the Review question: "Neither rp-cli nor codex detected. Install one for review support."

Step 7: Process Answers

Only process answers for questions that were asked (config values that were unset). Skip processing for config that was already set.

Memory (if question was asked):

• If "Yes": "${PLUGIN_ROOT}/scripts/flowctl" config set memory.enabled true --json
• If "No": "${PLUGIN_ROOT}/scripts/flowctl" config set memory.enabled false --json

Plan-Sync (if question was asked):

• If "Yes": "${PLUGIN_ROOT}/scripts/flowctl" config set planSync.enabled true --json
• If "No": "${PLUGIN_ROOT}/scripts/flowctl" config set planSync.enabled false --json

Plan-Sync cross-epic (if question was asked):

• If "Yes": "${PLUGIN_ROOT}/scripts/flowctl" config set planSync.crossEpic true --json
• If "No": "${PLUGIN_ROOT}/scripts/flowctl" config set planSync.crossEpic false --json

Review (if question was asked): Map user's answer to config value and persist:

# Determine backend from answer
case "$review_answer" in
  "Codex"*) REVIEW_BACKEND="codex" ;;
  "RepoPrompt"*) REVIEW_BACKEND="rp" ;;
  *) REVIEW_BACKEND="none" ;;
esac

"${PLUGIN_ROOT}/scripts/flowctl" config set review.backend "$REVIEW_BACKEND" --json

Docs: For each chosen file (CLAUDE.md and/or AGENTS.md):

1. Read the file (create if doesn't exist)
2. If marker exists: replace everything between <!-- BEGIN FLOW-NEXT --> and <!-- END FLOW-NEXT --> (inclusive)
3. If no marker: append the snippet from templates/claude-md-snippet.md

Star:

• If "Yes, star it":
  1. Check if gh CLI is available: which gh
  2. If available, run: gh api -X PUT /user/starred/gmickel/gmickel-claude-marketplace
  3. If gh not available or command fails, show: Star manually: https://github.com/gmickel/gmickel-claude-marketplace

Step 8: Print Summary

Flow-Next setup complete!

Installed:
- .flow/bin/flowctl (v<VERSION>)
- .flow/bin/flowctl.py
- .flow/usage.md

To use from command line:
  export PATH=".flow/bin:$PATH"
  flowctl --help

Configuration (use flowctl config set to change):
- Memory: <enabled|disabled>
- Plan-Sync: <enabled|disabled>
- Plan-Sync cross-epic: <enabled|disabled>
- Review backend: <codex|rp|none>

Documentation updated:
- <files updated or "none">

Notes:
- Re-run /flow-next:setup after plugin updates to refresh scripts
- Interested in autonomous mode? Run /flow-next:ralph-init
- Uninstall (run manually): rm -rf .flow/bin .flow/usage.md and remove <!-- BEGIN/END FLOW-NEXT --> block from docs
- This setup is optional - plugin works without it