Agent-Skills.md

Agent Skills: MUX Subagent Protocol

MUX data-plane worker protocol reference. Defines the report, signal, and return-code contract that strict mux coordinators consume.

UncategorizedID: MatiasComercio/agentic-config/mux-subagent

Repository

MatiasComercio/agentic-config
WaterplanAILicense: MIT
316

Install this agent skill to your local

pnpm dlx add-skill https://github.com/WaterplanAI/agentic-config/tree/HEAD/plugins/ac-workflow/skills/mux-subagent

Skill Files

Browse the full folder contents for mux-subagent.

Download Skill

Loading file tree…

plugins/ac-workflow/skills/mux-subagent/SKILL.md

Skill Metadata

Name
mux-subagent
Description
"MUX data-plane worker protocol reference. Defines the report, signal, and return-code contract that strict mux coordinators consume."

MUX Subagent Protocol

Purpose

This is the bounded leaf-worker contract for the mux family.

Binding activation

If the parent coordinator explicitly invokes mux-subagent, or embeds this skill text as the worker runtime for the task, treat this document as a binding runtime contract, not as optional guidance, commentary, or a planning reference.

Workers stay data-plane only. Control-plane ownership (LOCK/RESOLVE/DECLARE/DISPATCH/VERIFY/ADVANCE/BLOCK/RECOVER) stays with the coordinator plus shared mux tools.

Companion protocol artifacts:

  • ${CLAUDE_PLUGIN_ROOT}/mux/protocol/subagent.md
  • ${CLAUDE_PLUGIN_ROOT}/mux/protocol/guardrail-policy.md
  • ${CLAUDE_PLUGIN_ROOT}/mux/protocol/strict-happy-path-transcript.md
  • ${CLAUDE_PLUGIN_ROOT}/mux/protocol/strict-blocker-path-transcript.md
  • ${CLAUDE_PLUGIN_ROOT}/mux/protocol/strict-regression-checklist.md

Return Code Convention

Your final textual response to the parent coordinator must be exactly: 0

  • 0 = success
  • Any other final response = protocol violation
  • All substantive content belongs in the report file

Mandatory Worker Rules

  • Write all substantive findings and results to the report file path provided by the coordinator.
  • Create a success or failure signal before finishing.
  • Keep final response text exactly 0 on success.
  • Do not launch nested subagent calls.
  • Do not call control-plane bridge tools or report_parent; worker-to-coordinator communication is report/signal artifacts only.
  • Do not perform control-plane transitions from a worker.
  • Do not use TaskOutput.

Signal Command

uv run ${CLAUDE_PLUGIN_ROOT}/mux/tools/signal.py <signal-path> --path <report-path> --status success

Report Format

# <Report Title>

## Table of Contents
(all markdown headers)

## Executive Summary
- **Finding/Result**: keyword-dense value
- **Finding/Result**: keyword-dense value
- (max 5-7 bullets)

### Next Steps
- **Recommended action**: what the orchestrator should do next
- **Dependencies**: files or reports this connects to
- **Routing hint**: which worker type should consume this next

## <Sections>
(your detailed content)

Executive Summary Rules

  • Bullets only — no paragraphs.
  • Each bullet uses **Label**: keyword-dense value.
  • Include concrete file paths, counts, statuses, and findings.
  • Never be vague when a concrete count or path is available.

Next Steps Rules

  • Always include the subsection.
  • Tell the coordinator what to do next.
  • Reference specific file paths the next worker should read.
  • Flag blockers or decisions that need user input.

Pre-Return Checklist

Before returning 0, verify:

  • [ ] Report written to the specified file path
  • [ ] Report has an Executive Summary section with bullets only
  • [ ] Executive Summary has a ### Next Steps subsection
  • [ ] Signal file created via ${CLAUDE_PLUGIN_ROOT}/mux/tools/signal.py
  • [ ] No nested subagent calls were made
  • [ ] No control-plane bridge tools or report_parent calls were made
  • [ ] No substantive content appears in the final response