Skip to content

rfe.speedrun

Execute the full RFE pipeline end-to-end -- create, auto-fix (review + revise + split), and submit -- with minimal interaction. Detects one of three modes: batch YAML input (multiple ideas in a file), one or more existing Jira keys, or a single free-text idea. In batch mode it validates the input file with validate_batch_input.py --strict before spending any agent budget, pre-allocates all RFE IDs, and launches one parallel rfe.create agent per entry. It always passes an explicit --batch-size to auto-fix for reproducibility, then verifies completeness with check_autofix_complete.py and re-invokes auto-fix on any missing IDs (up to 3 retries). Orchestrates purely by invoking rfe.create, rfe.auto-fix, and rfe.submit, persisting config and IDs to disk between phases to survive context compression.

Plugin: rfe-creator | User-invocable

Contract

Skill Contract canonical-skill-v1

Run the full RFE pipeline end-to-end — create, auto-fix, and submit — from an idea, Jira key(s), or a batch file.

Identity

Functions
orchestrate
Success
  • Runs create, auto-fix, and submit in sequence, persisting IDs and config between phases.
  • Verifies completeness and retries missing IDs before submitting.

Optimization Targets

task_success judge SKILL.md @ c8a2a1e

Invariants

Must Preserve
  • Always pass an explicit batch size to auto-fix for reproducibility.
  • Never emit a text-only response during pipeline execution, which terminates the CI process.
Fixed Context
toolsRead, Write, Edit, Glob, Grep, Bash, AskUserQuestion, Skill
clipython3
knowledgerepository_contentpublic, task_inputtask_private, tool_outputtask_private

Traceability

Skill
.claude/skills/rfe.speedrun/SKILL.md

Diagram

rfe.speedrun diagram

Arguments

/rfe.speedrun <idea-or-key> [--input <path>] [--headless] [--dry-run] [--batch-size N] [--announce-complete]
Argument Required Default Description
idea-or-key - A free-text idea or Jira key (RHAIRFE-NNNN). Mutually exclusive with --input.
--input - Path to a YAML file with batch entries (prompt, priority, labels per entry)
--headless - Suppress questions and confirmations (for CI/eval)
--dry-run - Skip Jira writes in submit phase
--batch-size 5 Override batch size for auto-fix phase (always passed through explicitly)
--announce-complete - Print completion marker when done (for CI/eval harnesses)

Usage

/rfe.speedrun Better dashboard for ML model monitoring
/rfe.speedrun RHAIRFE-1234
/rfe.speedrun --input batch.yaml --headless --dry-run
/rfe.speedrun --input batch.yaml --headless --batch-size 10 --announce-complete