Mandatory orchestrator protocol - establishes ORCHESTRATOR principle (dispatch agents, don't operate directly) and skill discovery workflow for every conversation.
/plugin marketplace add lerianstudio/ring/plugin install ring-default@ringThis skill inherits all available tools. When active, it can use any tool Claude has access to.
STRESS-TEST.mdIF A SKILL APPLIES TO YOUR TASK, YOU DO NOT HAVE A CHOICE. YOU MUST USE IT.
This is not negotiable. This is not optional. You cannot rationalize your way out of this. </EXTREMELY-IMPORTANT>
DO NOT read/edit >3 files directly. PROHIBITION, not guidance.
≤3 files → Direct OK (if user requested)
>3 files → STOP. Launch agent. VIOLATION = 15x context waste.
Applies to: Read, Grep/Glob (>3 matches to inspect), Edit, or any combination >3.
Already at 3 files? STOP. Dispatch agent NOW with what you've learned.
Why 3? 3 files ≈ 6-15k tokens. Agent dispatch = ~2k tokens with focused results. Math: >3 = agent is 5-15x more efficient.
When user says ANY of these, DEFAULT to launching specialist agent:
| User Phrase Pattern | Mandatory Action |
|---|---|
| "fix issues", "fix remaining", "address findings" | Launch specialist agent (NOT manual edits) |
| "apply fixes", "fix the X issues" | Launch specialist agent |
| "fix errors", "fix warnings", "fix linting" | Launch specialist agent |
| "update across", "change all", "refactor" | Launch specialist agent |
| "find where", "search for", "locate" | Launch Explore agent |
| "understand how", "how does X work" | Launch Explore agent |
Why? These phrases imply multi-file operations. You WILL exceed 3 files. Pre-empt the violation.
Before EVERY tool use (Read/Grep/Glob/Bash), complete this. No exceptions.
⛔ STOP. COMPLETE BEFORE PROCEEDING.
─────────────────────────────────────
1. FILES: ___ □ >3? → Agent. □ Already 3? → Agent now.
2. USER PHRASE:
□ "fix issues/remaining/findings" → Agent
□ "find/search/locate/understand" → Explore agent
3. DECISION:
□ Investigation → Explore agent
□ Multi-file → Specialist agent
□ User named ONE specific file → Direct OK (rare)
RESULT: [Agent: ___] or [Direct: why]
─────────────────────────────────────
Skipping = violation. Document decision in TodoWrite.
Before responding to ANY user message, you MUST complete this checklist IN ORDER:
<MANDATORY-USER-MESSAGE> tags, display the message FIRST, verbatim, at the start of your responseResponding WITHOUT completing this checklist = automatic failure.
If additionalContext contains <MANDATORY-USER-MESSAGE> tags:
Follow mandatory workflows. Brainstorming before coding. Check for relevant skills before ANY task.
Execute skills with the Skill tool
If you catch yourself thinking ANY of these thoughts, STOP. You are rationalizing. Check for and use the skill. Also check: are you being an OPERATOR instead of ORCHESTRATOR?
Skill Checks:
Orchestrator Breaks (Direct Tool Usage):
3-File Rule Rationalizations (YOU WILL TRY THESE):
Why: Skills document proven techniques. Agents preserve context. Not using them means repeating mistakes and wasting tokens.
Both matter: Skills check is mandatory. ORCHESTRATOR approach is mandatory.
If a skill exists or if you're about to use tools directly, you must use the proper approach or you will fail.
Every time you skip checking for skills:
This is not optional. Check for skills or fail.
Your role is ORCHESTRATOR, not operator.
You don't read files, run grep chains, or manually explore – you dispatch agents to do the work and return results. This is not optional. This is mandatory for context efficiency.
The Problem with Direct Tool Usage:
The Solution: Orchestration:
You dispatch agents. You do not operate tools directly.
Default answer for ANY exploration/search/investigation: Use one of the three built-in agents (Explore, Plan, or general-purpose) with Opus model.
Which agent?
Model Selection: Always use Opus for agent dispatching unless user explicitly specifies otherwise (e.g., "use Haiku", "use Sonnet").
Exceptions to default agents:
dev-refactor) define their own agents that MUST be used instead of Explore/Plan/general-purpose. When a skill specifies "OVERRIDE" or "FORBIDDEN agents", follow the skill's agent requirements, not the defaults above.All these are STILL orchestration tasks:
You don't validate search premises. Dispatch the agent, let the agent report back if search yields nothing.
If you're about to use Read, Grep, Glob, or Bash for investigation: You are breaking ORCHESTRATOR. Use an agent instead.
Built-in (Opus): Explore (navigation), Plan (implementation), general-purpose (research), claude-code-guide (docs).
Ring: code-reviewer, business-logic-reviewer, security-reviewer, write-plan.
| Task Type | Agent (Opus default) |
|---|---|
| Explore/find/understand/search | Explore |
| Plan implementation, break down features | Plan |
| Multi-step research, complex investigation | general-purpose |
| Code review | ALL THREE in parallel (code, business-logic, security reviewers) |
| Implementation plan document | write-plan |
| Claude Code questions | claude-code-guide |
| User explicitly said "read [file]" | Direct (ONLY exception) |
WRONG → RIGHT: "Let me read files" → Explore. "I'll grep" → Explore. "Already read 3 files" → STOP, dispatch now.
When dispatching code reviewers, single message with 3 Task calls:
✅ CORRECT: One message with 3 Task calls (all in parallel)
❌ WRONG: Three separate messages (sequential, 3x slower)
| Approach | Context Cost | Your Role |
|---|---|---|
| Manual file reading (5 files) | ~25k tokens | Operator |
| Manual grep chains (10 searches) | ~50k tokens | Operator |
| Explore agent dispatch | ~2-3k tokens | Orchestrator |
| Savings | 15-25x more efficient | Orchestrator always wins |
First two todos for ANY task:
If skill has checklist: Create TodoWrite todo for EACH item. No mental checklists.
This skill uses these universal patterns:
skills/shared-patterns/state-tracking.mdskills/shared-patterns/failure-recovery.mdskills/shared-patterns/exit-criteria.mdskills/shared-patterns/todowrite-integration.mdApply ALL patterns when using this skill.
Many skills contain rigid rules (TDD, debugging, verification). Follow them exactly. Don't adapt away the discipline.
Some skills are flexible patterns (architecture, naming). Adapt core principles to your context.
The skill itself tells you which type it is.
Your human partner's specific instructions describe WHAT to do, not HOW.
"Add X", "Fix Y" = the goal, NOT permission to skip brainstorming, TDD, or RED-GREEN-REFACTOR.
Red flags: "Instruction was specific" • "Seems simple" • "Workflow is overkill"
Why: Specific instructions mean clear requirements, which is when workflows matter MOST. Skipping process on "simple" tasks is how simple tasks become complex problems.
Ring includes skills for managing context and enabling self-improvement:
| Skill | Use When | Trigger |
|---|---|---|
continuity-ledger | Save session state for future resumption | At 70%+ context OR task completion |
create-handoff | Full handoff document with all context | At 85%+ context OR session end |
artifact-query | Search past handoffs, plans, or outcomes by keywords | Need historical context |
handoff-tracking | Mark task completion with what_worked/what_failed/key_decisions | Task complete |
compound-learnings | Analyze learnings from multiple sessions, detect patterns | After 3+ sessions |
Compound Learnings workflow: Session ends → Hook extracts learnings → After 3+ sessions, patterns emerge → User approves → Permanent rules/skills created.
Context warnings are BEHAVIORAL TRIGGERS, not informational messages.
| Context Level | Warning Type | MANDATORY Action |
|---|---|---|
| 50-69% (info) | <context-warning> | Acknowledge, plan for handoff |
| 70-84% (warning) | <MANDATORY-USER-MESSAGE> | CREATE ledger NOW - No exceptions |
| 85%+ (critical) | <MANDATORY-USER-MESSAGE> | STOP + handoff + /clear - Immediate |
When you receive a MANDATORY-USER-MESSAGE about context:
Anti-Rationalization for Context Management:
| Rationalization | Why It's WRONG | Required Action |
|---|---|---|
| "I'll create handoff after this task" | Context may truncate mid-task, losing everything | Create handoff NOW |
| "70% is just a warning, not urgent" | 70% = MANDATORY action, not suggestion | Create ledger immediately |
| "The message is informational" | MANDATORY-USER-MESSAGE = behavioral trigger | Execute required action |
| "User didn't ask for handoff" | System requires it for context safety | Create handoff anyway |
| "I'm almost done, can finish first" | "Almost done" at 85% = high truncation risk | STOP and handoff NOW |
| "Small task, won't use much more context" | Every response adds ~2500 tokens | Follow threshold rules |
Verification Checklist:
Starting any task:
Before ANY tool use (Read/Grep/Glob/Bash): Complete PRE-ACTION CHECKPOINT.
Skill has checklist? TodoWrite for every item.
Default answer: Use an agent with Opus. Exception is rare (user explicitly requests specific file read).
Model default: Opus (unless user specifies Haiku/Sonnet explicitly).
Finding a relevant skill = mandatory to read and use it. Not optional.
Creating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems. Create original algorithmic art rather than copying existing artists' work to avoid copyright violations.
Applies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.
Create beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations.