with one click
init
Initialize an iterative research project with spec and progress tracking.
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.
Menu
Initialize an iterative research project with spec and progress tracking.
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
Capture session learnings and save to skills, guidelines, or reference docs under ~/.claude/.
Orchestrate parallel claude -p sessions — bootstrap, launch, monitor, and converge. Works with any skill that produces manifest.json, item directories, and a runner script.
Create a request (pull request or merge request) or update an existing one following project conventions.
Resolve merge or rebase conflicts between branches.
Assess open PRs with unaddressed review comments and generate a parallel addressing script — produces manifest.json and let-it-rip.sh for address-request-comments execution.
Assess open work items and generate a parallel execution script — produces manifest.json and let-it-rip.sh for implement/clarify execution.
| name | init |
| description | Initialize an iterative research project with spec and progress tracking. |
| disable-model-invocation | true |
| argument-hint | [topic] |
| allowed-tools | ["Read","Write","Bash"] |
Create a new Ralph loop project in an isolated git worktree with customized spec and progress files.
/ralph:init <topic> - Create project for the given topic/ralph:init - Will prompt for topicRead these templates from ~/.claude/ralph/research/templates/ before creating project files:
spec-template.md - Template for spec.md with v2 features (dynamic tasks, deep research workflow)progress-template.md - Template for progress.md with questions sectionGet project topic:
$ARGUMENTS provided, use that as the topicDerive project name:
Check for existing worktree:
claude/worktrees/research-<project-name>/ already exists, offer three options:
git worktree remove then create freshCreate worktree:
git worktree add claude/worktrees/research-<project-name> -b research/<project-name> HEAD
research/<project-name> already exists, use it without -b:
git worktree add claude/worktrees/research-<project-name> research/<project-name>
Create project directory inside worktree:
mkdir -p claude/worktrees/research-<project-name>/docs/staged-learnings/<project-name>
Create spec.md inside worktree using @spec-template.md:
claude/worktrees/research-<project-name>/docs/staged-learnings/<project-name>/spec.md<PROJECT_NAME> with the topic (title case)<TOPIC> with the topic<output_file> with info.mdCreate progress.md inside worktree using @progress-template.md:
claude/worktrees/research-<project-name>/docs/staged-learnings/<project-name>/progress.md<TOPIC> with the topicConfirm to the operator:
Created Ralph research project in worktree.
Worktree: claude/worktrees/research-<project-name>/
Branch: research/<project-name>
Project: docs/staged-learnings/<project-name>/
Next steps:
cd claude/worktrees/research-<project-name>
bash ~/.claude/ralph/research/wiggum.sh docs/staged-learnings/<project-name>
/ralph:init options pricing models
Created Ralph research project in worktree.
Worktree: claude/worktrees/research-options-pricing-models/
Branch: research/options-pricing-models
Project: docs/staged-learnings/options-pricing-models/
Next steps:
cd claude/worktrees/research-options-pricing-models
bash ~/.claude/ralph/research/wiggum.sh docs/staged-learnings/options-pricing-models
The "Areas for Deeper Investigation" section in info.md serves as the tracker for research progress. Use strikethrough + links to show completed investigations:
## Areas for Deeper Investigation
1. ~~Token optimization strategies~~ → See [token-optimization.md](./token-optimization.md)
2. ~~CI/CD integration~~ → See [ci-cd-integration.md](./ci-cd-integration.md)
3. Quality metrics - measuring output quality across iterations
4. Context handoff patterns - efficient state transfer between iterations
Why this format:
Alternative for lighter investigations (no separate file produced):
- [x] Token optimization strategies
- [ ] Quality metrics
Create separate <topic>.md files when:
Append to info.md when:
The hybrid approach keeps info.md lean as a navigation hub while substantial research lives in dedicated files.
When a project directory already exists, offer three options — not just overwrite/abort:
-1b suffix (e.g., topic-1b/) for parallel comparisonUse -1b (not -v2) because parallel research is for comparison, not versioning. If -1b exists, increment: -1c, -1d, etc.