agentleFS
kreighmitchellkreighmitchell/SuperClaude/.claude/skills/sc-mcp/SKILL.md

sc-mcp

Skill4 starsChanged 8 months agoMITView on GitHub

Comprehensive MCP orchestration skill integrating PAL MCP (reasoning, consensus, debugging) and Rube MCP (500+ app automations). Central hub for all MCP-powered workflows.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 6 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Discover: - Use RUBE_SEARCH_TOOLS for external needs, listmodels for PAL

    line 186
  2. Very specific:

    Plan: - Create execution plan (PAL planner or RUBE_CREATE_PLAN)

    line 187
  3. Distinctive:

    Validate: - Use consensus for critical decisions

    line 188
  4. Distinctive:

    Execute: - Run PAL analysis and/or Rube tools

    line 189
  5. Very specific:

    Use continuation_id to maintain context across PAL tool calls:

    line 197
  6. Very specific:

    Use session_id and memory for Rube continuity:

    line 210
  7. Distinctive:

    Always search tools before executing unknown integrations

    line 252
  8. Distinctive:

    Use consensus for decisions with >$1000 impact

    line 253
  9. Distinctive:

    Check connection status before automation

    line 256
  10. Distinctive:

    Use thinking_mode=high for complex PAL analysis

    line 257

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

MCP Orchestration Skill

Central orchestration hub for PAL MCP and Rube MCP capabilities. Use this skill for complex workflows requiring multi-model reasoning, external service integration, or both.

Quick Start

# PAL-powered analysis
/sc:mcp analyze --pal consensus --question "Should we use microservices?"

# Rube-powered automation
/sc:mcp automate --rube --apps slack,github --workflow "notify on PR"

# Combined orchestration
/sc:mcp orchestrate --pal thinkdeep --rube --full-validation

PAL MCP Integration

Available Tools

ToolInvocationPurpose
chatmcp__pal__chatCollaborative thinking, brainstorming
thinkdeepmcp__pal__thinkdeepMulti-stage investigation, complex analysis
plannermcp__pal__plannerSequential planning with branching
consensusmcp__pal__consensusMulti-model voting on decisions
codereviewmcp__pal__codereviewSystematic code quality analysis
precommitmcp__pal__precommitGit change validation
debugmcp__pal__debugRoot cause analysis
challengemcp__pal__challengeForce critical thinking
apilookupmcp__pal__apilookupCurrent API/SDK documentation
listmodelsmcp__pal__listmodelsAvailable AI models
clinkmcp__pal__clinkExternal CLI integration

PAL Workflows

Consensus Decision Making

Use consensus for:
- Architectural decisions (2-3 models)
- Security validations (security-focused models)
- Technology choices (diverse perspectives)
- Complex trade-off analysis

Recommended model combinations:

  • Architectural: gpt-5.2 (for), gemini-3-pro (against), deepseek (neutral)
  • Security: gpt-5.2 (security focus), gemini-3-pro (attack surface)
  • Performance: gpt-5.2 (optimization), deepseek (efficiency)

Debug Investigation

Use debug for:
- Complex bugs with unclear causes
- Performance issues
- Race conditions
- Memory leaks
- Integration problems

Debug confidence levels: exploring -> low -> medium -> high -> very_high -> almost_certain -> certain

Code Review

Use codereview for:
- Pre-merge validation
- Security audits
- Performance reviews
- Architecture compliance

Review types: full, security, performance, quick

Rube MCP Integration

Available Tools

ToolInvocationPurpose
SEARCH_TOOLSmcp__rube__RUBE_SEARCH_TOOLSDiscover available integrations
GET_RECIPE_DETAILSmcp__rube__RUBE_GET_RECIPE_DETAILSGet details of saved recipes
MULTI_EXECUTEmcp__rube__RUBE_MULTI_EXECUTE_TOOLParallel tool execution
REMOTE_BASHmcp__rube__RUBE_REMOTE_BASH_TOOLRemote shell commands
REMOTE_WORKBENCHmcp__rube__RUBE_REMOTE_WORKBENCHPython sandbox execution
CREATE_RECIPEmcp__rube__RUBE_CREATE_UPDATE_RECIPESave reusable workflows
EXECUTE_RECIPEmcp__rube__RUBE_EXECUTE_RECIPERun saved recipes
FIND_RECIPEmcp__rube__RUBE_FIND_RECIPESearch existing recipes
MANAGE_CONNECTIONSmcp__rube__RUBE_MANAGE_CONNECTIONSApp authentication
GET_SCHEMASmcp__rube__RUBE_GET_TOOL_SCHEMASTool input schemas
MANAGE_SCHEDULEmcp__rube__RUBE_MANAGE_RECIPE_SCHEDULERecipe scheduling

Rube Workflows

External Integration Flow

1. SEARCH_TOOLS - Find relevant tools for use case
2. GET_SCHEMAS - Get input requirements (if schemaRef returned)
3. MANAGE_CONNECTIONS - Verify/create auth
4. MULTI_EXECUTE - Execute tools
5. CREATE_RECIPE - Save for reuse (optional)

Bulk Processing Flow

1. SEARCH_TOOLS - Find data source/destination tools
2. REMOTE_WORKBENCH - Process with Python helpers:
   - run_composio_tool() - Execute Composio tools
   - invoke_llm() - AI processing
   - upload_local_file() - Export results
   - proxy_execute() - Direct API calls

Supported Apps (500+)

Communication: Slack, Discord, Teams, Gmail, Outlook, WhatsApp, Telegram Development: GitHub, GitLab, Jira, Linear, Asana, Vercel Productivity: Google Workspace, Notion, Airtable, Trello Data: Snowflake, BigQuery, Datadog, Amplitude AI: OpenAI, Anthropic, Replicate

Combined Orchestration Patterns

Pattern 1: Research + Decide + Execute

1. PAL thinkdeep - Investigate problem deeply
2. PAL consensus - Get multi-model decision
3. Rube SEARCH_TOOLS - Find execution tools
4. Rube MULTI_EXECUTE - Implement decision

Pattern 2: Review + Validate + Notify

1. PAL codereview - Review code changes
2. PAL precommit - Validate git changes
3. Rube MULTI_EXECUTE - Send notifications (Slack, email)
4. Rube CREATE_RECIPE - Save for CI/CD

Pattern 3: Debug + Fix + Verify

1. PAL debug - Root cause analysis
2. Implement fix locally
3. PAL codereview - Validate fix
4. Rube MULTI_EXECUTE - Update tickets, notify team

Pattern 4: Plan + Consensus + Automate

1. PAL planner - Create implementation plan
2. PAL consensus - Validate approach with multiple models
3. Rube MULTI_EXECUTE - Execute across apps
4. Rube MULTI_EXECUTE - Execute across apps
5. Rube CREATE_RECIPE - Save as reusable workflow

Flags

FlagTypeDefaultDescription
--palstring-PAL tool: chat, thinkdeep, planner, consensus, codereview, precommit, debug
--rubeboolfalseEnable Rube MCP integration
--appsstring-Comma-separated apps for Rube
--modelsstringautoModels for consensus (comma-separated)
--full-validationboolfalseRun all PAL validators
--save-recipeboolfalseSave workflow as Rube recipe
--schedulestring-Cron expression for recipe scheduling

Behavioral Flow

  1. Analyze - Understand what MCP capabilities are needed
  2. Discover - Use RUBE_SEARCH_TOOLS for external needs, listmodels for PAL
  3. Plan - Create execution plan (PAL planner or RUBE_CREATE_PLAN)
  4. Validate - Use consensus for critical decisions
  5. Execute - Run PAL analysis and/or Rube tools
  6. Persist - Save recipes, store memory for continuity
  7. Report - Present findings with tool attribution

Memory & State Management

PAL Continuation

Use continuation_id to maintain context across PAL tool calls:

# First call returns continuation_id
result = mcp__pal__thinkdeep(...)
continuation_id = result["continuation_id"]

# Subsequent calls reuse it
result = mcp__pal__thinkdeep(..., continuation_id=continuation_id)

Rube Session & Memory

Use session_id and memory for Rube continuity:

# First search generates session_id
result = mcp__rube__RUBE_SEARCH_TOOLS(..., session={"generate_id": True})
session_id = result["session_id"]

# Subsequent calls reuse session and build memory
result = mcp__rube__RUBE_MULTI_EXECUTE_TOOL(
    ...,
    session_id=session_id,
    memory={"slack": ["Channel general is C123"]}
)

Examples

Multi-Model Architecture Review

/sc:mcp analyze --pal consensus --models "gpt-5.2,gemini-3-pro,deepseek" \
  --question "Is event sourcing appropriate for this use case?"

Automated PR Workflow

/sc:mcp automate --rube --apps github,slack \
  --workflow "On PR merge, post summary to #releases"
  --save-recipe --schedule "0 9 * * 1-5"

Full Investigation Pipeline

/sc:mcp orchestrate --pal debug --rube \
  --issue "Memory leak in production" \
  --notify slack,jira --full-validation

Guardrails

  • Always search tools before executing unknown integrations
  • Use consensus for decisions with >$1000 impact
  • Validate schemas before multi-execute
  • Store memory for frequently used IDs
  • Check connection status before automation
  • Use thinking_mode=high for complex PAL analysis

Error Handling

ErrorRecovery
PAL model unavailableFall back to different model
Rube connection missingPrompt MANAGE_CONNECTIONS
Tool schema unknownCall GET_SCHEMAS first
Rate limitedUse backoff in REMOTE_WORKBENCH
Recipe not foundSearch or create new

Resources

  • PAL MCP: codereview, debug, consensus, thinkdeep, precommit, planner, chat, challenge, apilookup
  • Rube MCP: 500+ app integrations via Composio
  • Trait: mcp-pal-enabled - Apply PAL to any agent
  • Trait: mcp-rube-enabled - Apply Rube to any agent

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.