Agent skill

ralph

Queue processing with fresh context per phase. Processes N tasks from the queue, spawning isolated subagents to prevent context contamination. Supports serial, parallel, batch filter, and dry run modes. Triggers on "/ralph", "/ralph N", "process queue", "run pipeline tasks".

majiayu000534★ · 1 repos on radarProfile →
claude-codecan modify filesMIT
Install
npx skills add majiayu000/claude-skill-registry --skill ralph-lopewale-dev-ignition --agent claude-code

Same command for any agent — swap --agent for codex, cursor, copilot.

Facts
Files in the skill folder: 2
SKILL.md size: 21 KB
Bundled scripts: none
Version: 1.0
Allowed tools: ReadWriteEditGrepGlobBashTask
Path: skills/agent/ralph-lopewale-dev-ignition/SKILL.md
Open the folder on GitHub →
Where it comes from
Stars: 534
Language: HTML

Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.

Review
written from the skill's own SKILL.md · Aug 5, 2026

What it does

Spawns a subagent for every task to process items pulled from a queue, ensuring isolated context per phase. Supports serial processing with optional parallel mode, and can filter by batch or by current phase type. Provides a dry-run option to preview actions without executing. Outputs a RALPH HANDOFF block at the end of each task to communicate work done and any queue updates.

How it works

  1. Parse arguments: N tasks to process (required unless --dry-run), optional --parallel, --batch, --type, --dry-run, and --handoff.
  2. Read queue state from ops/queue.yaml, ops/queue/queue.yaml, or ops/queue/queue.json to identify pending tasks.
  3. Build a list of actionable tasks (status == "pending"), applying --batch and --type filters if provided.
  4. If --dry-run, output a summary of the queue and next tasks, then stop.
  5. In serial mode (default unless --parallel): for up to N tasks, select the next pending task, read its metadata, and construct a phase-specific prompt including:
    • task file path, task id, current_phase, and target
    • instruction to use the skill with --handoff and ONE PHASE ONLY
    • mandated subagent spawning for every task via the Task tool
    • output of a RALPH HANDOFF block
  6. Spawn a subagent using Task with the constructed prompt; wait for completion.
  7. On return, extract the RALPH HANDOFF block if present, capture Learnings, and queue updates.
  8. Advance the task to the next phase (or mark done if last phase) and update current_phase and completed_phases accordingly.
  9. Re-filter the queue for the next iteration and repeat until N tasks are processed or none are pending.
  10. In parallel mode (--parallel): process up to 5 concurrent claims through all phases, with a separate Phase B cross-connect validation step after completion. Incompatibility with --type is enforced, returning an error if both are provided.

When to use it

Use when you need structured, isolated context per task phase and want automated, per-task subagent execution with explicit phase sequencing. Use --dry-run to validate impact before execution. Use --parallel to process multiple claims end-to-end concurrently, then validate cross-connections.

What it can touch

  • Task tool is mandatory for spawning subagents for every task.
  • Access to queue files: ops/queue.yaml, ops/queue/queue.yaml, ops/queue/queue.json for reading and updating task state.
  • Phase-specific prompts reference: ops/queue/{FILE} and {TARGET}, {ID}, {CURRENT_PHASE}, {DOMAIN:note} paths.

Caveats

  • Subagent spawning is mandatory for every task; inline execution is prohibited.
  • The lead session must not perform task work directly; only delegate via Task tool calls.
  • If a RALPH HANDOFF block is missing, the system will log a warning but proceed.
  • The final report must reflect phase progression logic according to the queue's phase_order; last phase marks done and timestamps appropriately.
From the SKILL.md

## EXECUTE NOW **Target: $ARGUMENTS** Parse arguments: - N (required unless --dry-run): number of tasks to process - --parallel: concurrent claim workers (max 5) + cross-connect validation - --batch [id]: process only tasks from specific batch - --type [type]: process only tasks at a specific phase (extract, create, reflect, reweave, verify, enrich) - --dry-run: show what would execute without running - --handoff: output structured RALPH HANDOFF block at end (for pipeline chaining) ### Step 0: Read Vocabulary Read `ops/derivation-manifest.md` (or fall back to `ops/derivation.md`) for domain vocabulary mapping. All output must use domain-native terms. If neither file exists, use universal terms. **START NOW.** Process queue tasks. --- ## MANDATORY CONSTRAINT: SUBAGENT SPAWNING IS NOT OPTIONAL **You MUST use the Task tool to spawn a subagent for EVERY task. No exceptions.** This is not a suggestion. This is not an optimization you can skip for "simple" tasks. The entire architecture depends on fresh context isolation per phase. Executing tasks inline in the lead session: - Contaminates context (later tasks run on degraded attention) - Skips the handoff protocol (learnings are not cap

What's inside
Steps it walks through
  1. EXECUTE NOW
  2. Step 0: Read Vocabulary
  3. MANDATORY CONSTRAINT: SUBAGENT SPAWNING IS NOT OPTIONAL
  4. Phase Configuration
  5. Step 1: Read Queue State
  6. Step 2: Filter Tasks
  7. Step 3: If --dry-run, Report and Stop
  8. Step 4: Process Loop (SERIAL MODE)
  9. 4a. Select Next Task
  10. 4b. Build Subagent Prompt
  11. 4c. Spawn Subagent (MANDATORY — NEVER SKIP)
  12. 4d. Evaluate Return
  13. 4e. Update Queue (Phase Progression)
  14. 4f. Report Progress
Ships with 1 file
  • metadata.json
More from claude-skill-registry
All skills →
About this skill
What does the ralph skill do?

Queue processing with fresh context per phase. Processes N tasks from the queue, spawning isolated subagents to prevent context contamination. Supports serial, parallel, batch filter, and dry run modes. Triggers on "/ralph", "/ralph N", "process queue", "run pipeline tasks".

How do I install it?

Run `npx skills add majiayu000/claude-skill-registry --skill ralph-lopewale-dev-ignition --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.

Where does this skill come from?

From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.

Is a popular skill a good skill?

Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.

Keep going