Menu
Use when you want to review code or plans with an external model (ChatGPT, Claude web, etc.). Triggers on /flow-code:export-context.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Internal pipeline engine. Manages the entire pipeline (brainstorm, plan, plan-review, work, impl-review, close) via flowctl phase commands. Invoked by /flow-code:go.
Use when exploring requirements before planning. Pressure-tests ideas, generates approaches, and outputs a requirements doc for /flow-code:plan. Supports --auto mode for AI self-interview (no human input needed).
Use when exploring requirements before planning. Pressure-tests ideas, generates approaches, and outputs a requirements doc for /flow-code:plan.
Use when planning features or designing implementation. Triggers on /flow-code:plan with text descriptions or Flow IDs.
Use when reviewing code changes — self-review in Worker Phase 6, impl-review, or PR review. Applies five-axis scoring with severity labels.
Use when implementing a plan or working through a spec. Triggers on /flow-code:work with Flow IDs.
| name | flow-code-export-context |
| description | Use when you want to review code or plans with an external model (ChatGPT, Claude web, etc.). Triggers on /flow-code:export-context. |
| tier | 1 |
Export flow-code context to a markdown file for external LLMs (ChatGPT Pro, Claude web, etc.).
Arguments: $ARGUMENTS — Format: <type> <target> [focus areas]
plan <epic-id> — Export plan review contextimpl — Export implementation review context (current branch)FLOWCTL="$HOME/.flow/bin/flowctl"
OUTPUT_FILE="prompt-exports/$(date +%Y%m%d-%H%M%S)-export.md"
mkdir -p prompt-exports
Plan: $FLOWCTL show <epic-id> --json, $FLOWCTL cat <epic-id>, gather task specs.
Impl: git branch --show-current, git log main..HEAD --oneline, git diff main..HEAD --stat.
Build instructions from gathered context. Extract the real task from the request — strip meta-framing about exporting.
Tier 1 — RP MCP (if mcp__RepoPrompt__context_builder available):
context_builder(instructions="<task><extracted task></task>\n<context><flowctl content></context>", response_type="clarify")
prompt(op="export", path="<OUTPUT_FILE>", copy_preset="<plan|codeReview>")
Tier 2 — rp-cli (if which rp-cli succeeds):
WINDOW_ID=$(rp-cli -e 'windows' | head -1 | awk '{print $1}')
rp-cli -w "$WINDOW_ID" -e 'builder "<instructions>" --response-type clarify'
rp-cli -w "$WINDOW_ID" -e "prompt export \"$OUTPUT_FILE\" --copy-preset <plan|codeReview>"
Tier 3 — Basic Markdown (no RP available):
Write gathered content directly to $OUTPUT_FILE as structured markdown with sections for context, specs, changed files, and focus areas.
Preset mapping: plan -> plan, impl -> codeReview.
Report $OUTPUT_FILE path, tier used, and export type. Instruct user to paste into their preferred external LLM.
Manual external review only. No Ralph support (no receipts, no status updates).