home / skills / simhacker / moollm / mooco-mirror

mooco-mirror skill

/skills/mooco-mirror

This skill compares MOOCO and Cursor sessions to reveal divergences, provenance, and optimization opportunities across orchestration decisions.

npx playbooks add skill simhacker/moollm --skill mooco-mirror

Review the files below or copy the command above to add this skill to your agents.

Files (5)
SKILL.md
1.9 KB
---
name: mooco-mirror
description: "Parallel introspection for MOOCO sessions and cross-orchestrator comparison"
license: MIT
tier: 0
allowed-tools: []
protocol: MOOCO-MIRROR
tags: [moollm, introspection, orchestration, traceability]
related: [mooco, cursor-mirror, session-log, event-logging, yaml-jazz, thoughtful-commitment, skill-snitch]
---

# MOOCO Mirror

> *"Watch two orchestration realities at once."*

MOOCO‑mirror is a reflection skill for MOOCO sessions. It is built to compare orchestration decisions, context assembly, and tool execution between MOOCO and Cursor environments. It also runs in production without a Cursor sidecar.

## Core Capabilities

- **Session introspection**: read MOOCO event logs and context bundles
- **Cross‑mirror**: line up a MOOCO run with a Cursor run
- **Divergence analysis**: highlight where context, tools, or outputs differ
- **Trace provenance**: show which steps are deterministic vs LLM‑driven
- **Optimization**: find bottlenecks and reduce orchestration friction
- **Debugging**: isolate drift, regressions, and tool misuse
- **Exploration**: probe alternative orchestration strategies safely
- **Play‑Learn‑Lift**: capture learnings and uplift them into skills
- **Context analysis**: inspect what was loaded and why
- **Reverse engineering**: infer Cursor and MOOCO orchestration patterns

## Cross‑Mirror Pattern

1. Capture a MOOCO session event stream.
2. Capture the parallel Cursor session via cursor‑mirror.
3. Align by task goal or timestamp.
4. Compare context, tools, and emitted events.

## Use Cases

- Validate MOOCO orchestration choices
- Debug unexpected tool selection
- Diagnose context drift across orchestrators
- Improve skill portability by exposing hidden assumptions
- Support thoughtful-commitment by exporting prompts, thoughts, context assembly, and cause/effect into commit and PR messages
- Run skill-snitch style scans to detect suspicious patterns in mirrored traces

Overview

This skill provides parallel introspection for MOOCO sessions and cross-orchestrator comparison. It helps teams observe, compare, and diagnose differences between MOOCO and Cursor orchestration runs. The tool surfaces provenance, divergence points, and optimization opportunities to improve portability and reliability.

How this skill works

The skill ingests MOOCO event logs and context bundles and can optionally pair them with Cursor session captures. It aligns traces by task goal or timestamps, then compares context assembly, tool selection, and emitted events. It highlights deterministic vs LLM-driven steps, flags divergence, and produces actionable diagnostics and exportable artifacts.

When to use it

  • Validating that MOOCO orchestration decisions match a Cursor reference run
  • Debugging unexpected tool selection or output differences across orchestrators
  • Diagnosing context drift or missing context elements between environments
  • Assessing portability of skills and exposing hidden assumptions
  • Generating commit-ready artifacts that explain orchestration cause/effect

Best practices

  • Capture full event streams and context bundles from both orchestrators for accurate alignment
  • Align runs by task goal first, then refine with timestamps to handle retries and async steps
  • Focus divergence analysis on tool inputs and context differences before blaming model outputs
  • Use exported artifacts (prompts, thoughts, context assembly) to create focused PR or test changes
  • Run periodic mirror scans to detect regressions or emerging suspicious patterns early

Example use cases

  • Compare a production MOOCO session to a Cursor-inspired local run to identify a missing context loader
  • Trace why a tool was chosen differently by showing the preceding context and thought steps
  • Audit orchestration changes by exporting paired traces into a commit message that documents cause and effect
  • Probe alternative orchestration strategies in a safe mirrored environment without affecting production
  • Run automated scans across mirrored traces to surface tool misuse or regressions

FAQ

Do I need a Cursor sidecar to use this skill?

No. It can run in production without a Cursor sidecar, though adding Cursor captures enables deeper cross-orchestrator comparison.

How does alignment handle retries or asynchronous steps?

Alignment starts by matching task goals, then refines with timestamps and event heuristics to correlate retries and async activity.