New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Ralph Specum

by tzachbon in tzachbon/smart-ralph

Use only when the user explicitly invokes $ralph-specum, requests Ralph Specum in Codex, or asks Ralph Specum to handle a named phase or optional prototype.

MITAuto-check passedDevelopment

Install Ralph Specum

skills CLI
$ npx skills add tzachbon/smart-ralph --skill ralph-specum -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install tzachbon/smart-ralph ralph-specum --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/tzachbon/smart-ralph.git skills-src && mkdir -p .claude/skills && cp -r skills-src/plugins/ralph-specum-codex/skills/ralph-specum .claude/skills/ralph-specum && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
ralph-specum
GitHub stars
559
Token cost
~2.4k tokens
SKILL.md length
1,119 words
Files
2
Skills in repo
26
Repo updated
First seen
Licence
MIT

At a glance

Use only when the user explicitly invokes $ralph-specum, requests Ralph Specum in Codex, or asks Ralph Specum to handle a named phase or optional prototype.

  • Works in 12 steps: You are a coordinator, not a doer.… → Keep the Ralph disk contract stable. → Treat .claude/ralph-specum.local.md as… → …
  • Explicitly invokes $ralph-specum
  • SKILL.md covers Read These References, Use These Helpers, Primary Routing and Core Rules, plus 4 more sections
  • Calls go

What it does

Ralph Specum is an agent skill from tzachbon/smart-ralph. Use only when the user explicitly invokes $ralph-specum, requests Ralph Specum in Codex, or asks Ralph Specum to handle a named phase or optional prototype.

Its SKILL.md is about 2.4k tokens, which your agent loads only when the skill is triggered. The skill folder holds 2 other files (for example `agents/openai.yaml`).

It sits in Development. The repository describes itself as: Spec-driven development with smart compaction. Claude Code plugin combining Ralph Wiggum loop with structured specification workflow. The licence is MIT.

When your agent uses it

  • Explicitly invokes $ralph-specum
  • Requests Ralph Specum in Codex
  • Asks Ralph Specum to handle a named phase
  • Optional prototype

Example prompts

  • “/ralph-specum”

Workflow steps

12 steps, taken from the first numbered list in SKILL.md.

  1. You are a coordinator, not a doer. Delegate each phase to the appropriate sub-agent and never write spec artifacts (research.md…
  2. Keep the Ralph disk contract stable.
  3. Treat .claude/ralph-specum.local.md as the settings source when present.
  4. Default to ./specs when no valid config exists.
  5. Keep .current-spec in the default specs root.
  6. Merge state fields. Do not replace the whole state object.
  7. Preserve source, name, goal, basePath, phase, taskIndex, totalTasks, taskIteration, maxTaskIterations, globalIteration…
  8. Also preserve newer state fields when present, especially awaitingApproval, quickMode, granularity, epicName, discoveredSkills, and native…
  9. Write .progress.md after every phase and after every implementation attempt.
  10. Keep pre-delegation interview approval separate from post-generation artifact approval.
  11. Honor the Commit line in tasks during implementation unless the user explicitly disables task commits.
  12. Use branch creation or worktree creation when the user asks for branch isolation or the repo policy requires it.

What it can do on your machine

Read from SKILL.md and the folder at commit ac7251a. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • go

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Ralph Specum loads about 2.4k tokens when it runs. Until then it costs about 43 tokens; SKILL.md has 1,119 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~43
When it runs · the whole SKILL.md, loaded when a task matches
~2.4k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from tzachbon/smart-ralph at commit ac7251a, republished under its MIT licence (© tzachbon). 1,119 words, ~2,412 tokens.

Download SKILL.mdSave it as .claude/skills/ralph-specum/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.
name
ralph-specum
description
Use only when the user explicitly invokes `$ralph-specum`, requests Ralph Specum in Codex, or asks Ralph Specum to handle a named phase or optional prototype.
metadata.surface
primary

Ralph Specum

Use this as the primary Codex surface for Ralph Specum. It carries the full reusable workflow and can handle the entire command surface directly when helper skills are not installed.

Derive RALPH_CODEX_PLUGIN_ROOT from this loaded skill: take the directory containing SKILL.md, then its parent skills directory, then the next parent. Never derive the plugin root from the project working directory.

Read These References

  • "$RALPH_CODEX_PLUGIN_ROOT/references/workflow.md" for the phase flow, branch and worktree behavior, quick mode, and command routing
  • "$RALPH_CODEX_PLUGIN_ROOT/references/state-contract.md" for .ralph-state.json, .progress.md, commit rules, and resume semantics
  • "$RALPH_CODEX_PLUGIN_ROOT/references/path-resolution.md" for specs_dirs, .current-spec, ambiguity handling, and default directory behavior
  • "$RALPH_CODEX_PLUGIN_ROOT/references/parity-matrix.md" for Claude-to-Codex feature translation and command mapping
  • "$RALPH_CODEX_PLUGIN_ROOT/references/prototype-coordinator.md" for direct, suggested, resume, quick, cancel, and prototype handoff behavior
  • "$RALPH_CODEX_PLUGIN_ROOT/skills/interview-framework-codex/SKILL.md" and its required algorithm and domain-modeling references for every normal-mode phase interview

Use These Helpers

  • "$RALPH_CODEX_PLUGIN_ROOT/scripts/resolve_spec_paths.py" for spec roots, current spec, and unique or ambiguous name resolution
  • "$RALPH_CODEX_PLUGIN_ROOT/scripts/merge_state.py" for safe top-level state merges
  • "$RALPH_CODEX_PLUGIN_ROOT/scripts/count_tasks.py" for task counts and next incomplete task
  • "$RALPH_CODEX_PLUGIN_ROOT/scripts/phase_gate.py" for mode, skill-load, interview, parent-delegation, and artifact-write gates
  • "$RALPH_CODEX_PLUGIN_ROOT/scripts/locked_state.py" for locked state and activePrototypes mutations
  • "$RALPH_CODEX_PLUGIN_ROOT/scripts/prototype_records.py" for reviewed immutable prototype records and downstream selection
  • "$RALPH_CODEX_PLUGIN_ROOT/scripts/prototype_harness.py" for bounded builder control outcomes and retry metadata
  • "$RALPH_CODEX_PLUGIN_ROOT/templates/" for the canonical Ralph markdown file shapes
  • "$RALPH_CODEX_PLUGIN_ROOT/assets/bootstrap/" when the user wants optional project-local Codex guidance

Primary Routing

Handle these intents directly:

IntentAction
Start, new, resume, quick modeFollow the start flow in ../../references/workflow.md
TriageDelegate to triage-analyst sub-agent to decompose into epic and specs
ResearchDelegate to research-analyst sub-agent to write research.md
RequirementsDelegate to product-manager sub-agent to write requirements.md
PrototypeFollow ../../references/prototype-coordinator.md, or route an explicit helper request to $ralph-specum-prototype
DesignDelegate to architect-reviewer sub-agent to write design.md
TasksDelegate to task-planner sub-agent to write tasks.md
ImplementDelegate each task to spec-executor sub-agent until complete or blocked
StatusShow active spec, backlog state, and per-root listing
SwitchUpdate .current-spec only
CancelStop execution and clean up state, confirm before destructive delete
IndexGenerate specs/.index/ component and external specs
RefactorDelegate to refactor-specialist sub-agent to update spec files
FeedbackOpen or draft GitHub feedback
HelpSummarize the surface and next commands

If the corresponding helper skill is installed and the user invoked it explicitly, keep behavior aligned with that helper. If not, perform the action here.

Core Rules

  1. You are a coordinator, not a doer. Delegate each phase to the appropriate sub-agent and never write spec artifacts (research.md, requirements.md, design.md, or tasks.md) yourself. Apply the shared hard-transition invariant to only start, triage, research, requirements, design, and tasks. For only start, triage, research, requirements, design, and tasks, enforce that invariant. For those phases, gather context, discover and preload contracts, run the interview, obtain final approval, pass the gate, delegate, validate the output, and present results for artifact approval. Keep the existing delegation flows for implement and refactor unchanged; the new phase gates do not apply to them.
  2. Keep the Ralph disk contract stable.
  3. Treat .claude/ralph-specum.local.md as the settings source when present.
  4. Default to ./specs when no valid config exists.
  5. Keep .current-spec in the default specs root.
  6. Merge state fields. Do not replace the whole state object.
  7. Preserve source, name, goal, basePath, phase, taskIndex, totalTasks, taskIteration, maxTaskIterations, globalIteration, maxGlobalIterations, commitSpec, and relatedSpecs.
  8. Also preserve newer state fields when present, especially awaitingApproval, quickMode, granularity, epicName, discoveredSkills, and native task sync metadata.
  9. Write .progress.md after every phase and after every implementation attempt.
  10. Keep pre-delegation interview approval separate from post-generation artifact approval.
  11. Honor the Commit line in tasks during implementation unless the user explicitly disables task commits.
  12. Use branch creation or worktree creation when the user asks for branch isolation or the repo policy requires it.
  13. Run phase_gate.py mode at entry to every affected phase. Only exact --quick enables quick mode; exact --interactive clears it. Reject both together, -q, variants, and natural-language substitutes. No flags normalize invalid legacy quick state to interactive.
  14. In exact quick mode, run discovery and contract loading, record the current manifest and bypassed_quick, generate missing artifacts, default task granularity to fine when unset, and continue into implementation in the same session.
  15. For normal-mode start, triage, research, requirements, design, and tasks, follow "$RALPH_CODEX_PLUGIN_ROOT/skills/interview-framework-codex/references/algorithm.md" before each delegation.
  16. Keep optional prototype work in activePrototypes; preserve the main phase and current checkout until the recorded handoff.
  17. Use child-agent controls and store only agentId for internal builders. Never use create_thread or threadId for them.
Show full SKILL.md (388 more words)Show less

Stop Enforcement

Before delegating any affected phase in normal mode, you MUST:

  1. Complete the applicable discovery pass and reload receipts
  2. Finish all critical decision frontiers
  3. Obtain explicit approve and delegate
  4. Pass phase_gate.py check-delegation

After completing any phase artifact (research, requirements, design, tasks), you MUST:

  1. Display the walkthrough summary
  2. Present the approval prompt (approve / request changes / continue to next)
  3. STOP and wait for user response

The ONLY exception is exact --quick mode with a valid bypass receipt. Without it, you MUST NOT auto-continue to the next phase.

Response Handoff

  • After writing research.md, requirements.md, design.md, tasks.md, or refactored spec files outside quick mode:
    • name the file or files that changed
    • give a short summary
    • end with exactly one explicit choice prompt:
      • approve current artifact
      • request changes
      • continue to <named next step>
  • Treat continue to <named next step> as approval of the current artifact and permission to enter that phase's discovery and interview path.
  • Treat apply the changes during artifact review as immediate authorization to delegate already-recorded revision feedback through a new unique dispatch. Redisplay and remain at artifact approval. Ask one focused change question only when no feedback is pending.
  • Treat control-only continue, proceed, and go ahead as approval of nothing.
  • After normal-mode start or new setup, run discovery pass 1 and begin the goal grill. Explicit final approval delegates research immediately.

Current Workflow Expectations

  • Use critical-decision frontier interviews for start, triage, research, requirements, design, and tasks when quick mode is not active. Ask no setup, administration, or discoverable question.
  • Route obviously large or cross-cutting efforts to triage before normal spec generation.
  • Support active epic state via specs/.current-epic and per-epic state in specs/_epics/<epic-name>/.
  • Treat task planning as POC-first with [P] markers for safe parallel work and [VERIFY] checkpoints for explicit quality validation.
  • Support VE tasks when the plan needs autonomous end-to-end verification.
  • During implementation, recompute task counts from disk, resume from the first incomplete task, and prefer task file truth over stale state.
  • Native task sync is part of the current Ralph execution model. Keep Codex wording aligned with that behavior without promising Claude-only hook mechanics.

Bootstrap

Bootstrap project-local files only when the user wants them.

Suggested bootstrap files:

  • "$RALPH_CODEX_PLUGIN_ROOT/assets/bootstrap/AGENTS.md" to give a consumer repo local Ralph guidance
  • "$RALPH_CODEX_PLUGIN_ROOT/assets/bootstrap/ralph-specum.local.md" to seed local settings

Do not bootstrap by default. Installation into $CODEX_HOME/skills is enough.

© tzachbon, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 1 other file in plugins/ralph-specum-codex/skills/ralph-specum of tzachbon/smart-ralph.

  • SKILL.md
  • agents/openai.yaml

Open the folder on GitHubat commit ac7251a

Compare with similar skills

Ralph Specum next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Ralph Specum compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Ralph Specum this skilltzachbon/smart-ralph559—~2.4kAutomated safety check: PassMIT
Vercel Composition Patternssupabase/supabase111k58 repos~726Automated safety check: PassMIT
Finishing a Development Branchobra/superpowers297k5 repos~1.9kAutomated safety check: PassMIT
Typescript Advanced Typesrolling-scopes/rsschool-app10k25 repos~4.2kAutomated safety check: PassMPL-2.0
PR Babysitteropeninterpreter/openinterpreter69k3 repos~4.2kAutomated safety check: PassApache-2.0
Code Review ChecklistshareAI-lab/learn-claude-code78k4 repos~1.1kAutomated safety check: PassMIT

Similar skills

  • Vercel Composition Patterns

    supabase/supabase

    Official

    React composition patterns that scale. An agent skill from supabase/supabase.

    111k GitHub starsUsed in 58 repos~726 tokens
    DevelopmentAuto-check passed

  • Finishing a Development Branch

    obra/superpowers

    Walks the last step of a branch: confirm tests pass, detect the git environment, ask how to integrate, carry out your choice and clean up the worktree.

    297k GitHub starsUsed in 5 repos~1.9k tokens
    DevelopmentAuto-check passed

  • Typescript Advanced Types

    rolling-scopes/rsschool-app

    Master TypeScript's advanced type system including generics, conditional types, mapped types, template literals, and utility types for building type-safe applications.

    10k GitHub starsUsed in 25 repos~4.2k tokens
    DevelopmentAuto-check passed

  • PR Babysitter

    openinterpreter/openinterpreter

    Watches an open GitHub pull request until it merges, handling review comments, diagnosing CI failures and retrying flaky checks along the way.

    69k GitHub starsUsed in 3 repos~4.2k tokens
    DevelopmentAuto-check passed

  • Code Review Checklist

    shareAI-lab/learn-claude-code

    Reviews code against a five-part checklist covering security, correctness, performance, maintainability and testing, and reports findings in a fixed format.

    78k GitHub starsUsed in 4 repos~1.1k tokens
    DevelopmentAuto-check passed

  • Greploop

    onyx-dot-app/onyx

    Iteratively improves a PR (GitHub), MR (GitLab), or shelved changelist (Perforce) until Greptile gives it a 5/5 confidence score with zero unresolved comments.

    32k GitHub starsUsed in 4 repos~3.3k tokens
    DevelopmentAuto-check passed

More from tzachbon/smart-ralph

All 26 skills in this repo

  • Communication Style

    tzachbon/smart-ralph

    This skill should be used when generating spec artifacts (research.md, requirements.md, design.md, tasks.md), formatting agent output, structuring phase results, or when any Ralph agent needs…

    559 GitHub stars~551 tokensUpdated 23 days ago
    Auto-check passed

  • Interview Framework

    tzachbon/smart-ralph

    This skill should be used when a Ralph phase must identify critical user decisions, run a layered grill, persist partial answers, obtain explicit approval, or resume an interrupted phase interview…

    559 GitHub stars~2.2k tokensUpdated 23 days ago
    Auto-check passed

  • Reality Verification

    tzachbon/smart-ralph

    This skill should be used when the user asks to "verify a fix", "reproduce failure", "diagnose issue", "check BEFORE/AFTER state", "VF task", "reality check", "check test quality", "mock-only…

    559 GitHub stars~877 tokensUpdated 23 days ago
    Auto-check passed

  • Spec Workflow

    tzachbon/smart-ralph

    This skill should be used when the user asks to "build a feature", "create a spec", "start spec-driven development", "run research phase", "generate requirements", "create design", "plan tasks"…

    559 GitHub stars~1.1k tokensUpdated 23 days ago
    Auto-check passed

  • Ralph Specum Design

    tzachbon/smart-ralph

    This skill should be used only when the user explicitly asks to use $ralph-specum-design, or explicitly asks Ralph Specum in Codex to run the design phase.

    559 GitHub stars~1.5k tokensUpdated 23 days ago
    Auto-check passed

  • Ralph Specum Tasks

    tzachbon/smart-ralph

    This skill should be used only when the user explicitly asks to use $ralph-specum-tasks, or explicitly asks Ralph Specum in Codex to run the tasks phase.

    559 GitHub stars~1.6k tokensUpdated 23 days ago
    Auto-check passed

Categories

Questions about Ralph Specum

What does Ralph Specum do?

Use only when the user explicitly invokes $ralph-specum, requests Ralph Specum in Codex, or asks Ralph Specum to handle a named phase or optional prototype. Ralph Specum is an agent skill from tzachbon/smart-ralph. Use only when the user explicitly invokes $ralph-specum, requests Ralph Specum in Codex, or asks Ralph Specum to handle a named phase or optional prototype.

When should I use Ralph Specum?

Ralph Specum fits situations like: explicitly invokes $ralph-specum; requests Ralph Specum in Codex; asks Ralph Specum to handle a named phase; optional prototype.

How do I install Ralph Specum in Claude Code?

Run `npx skills add tzachbon/smart-ralph --skill ralph-specum -a claude-code`. Or copy the skill folder (plugins/ralph-specum-codex/skills/ralph-specum in tzachbon/smart-ralph) into .claude/skills/ralph-specum in your project. Claude Code loads it when a task matches its description.

How do I install Ralph Specum in Codex?

Run `npx skills add tzachbon/smart-ralph --skill ralph-specum -a codex`. Or copy the skill folder (plugins/ralph-specum-codex/skills/ralph-specum in tzachbon/smart-ralph) into .agents/skills/ralph-specum in your project. Codex loads it when a task matches its description.

Can I use Ralph Specum in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add tzachbon/smart-ralph --skill ralph-specum -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/ralph-specum, .gemini/skills/ralph-specum, .github/skills/ralph-specum and .opencode/skills/ralph-specum in your project.

What does Ralph Specum need to run?

Going by SKILL.md and its folder, Ralph Specum needs the command-line tools its instructions call (go).

Does Ralph Specum access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Ralph Specum safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Ralph Specum use?

Ralph Specum is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Ralph Specum use?

About 2.4k tokens (SKILL.md is roughly 9.6k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Ralph Specum?

Skills that share tags, products or a category with Ralph Specum: Vercel Composition Patterns (supabase/supabase, 111k stars), Finishing a Development Branch (obra/superpowers, 297k stars), Typescript Advanced Types (rolling-scopes/rsschool-app, 10k stars) and PR Babysitter (openinterpreter/openinterpreter, 69k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Ralph Specum?

tzachbon (a GitHub user) maintains it in tzachbon/smart-ralph, which has 559 GitHub stars. The repository holds 26 skills in this directory. The repository was last updated on September 16, 2026.

Source: tzachbon/smart-ralph on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.