Claude-skill-registry agent-ops-guide

Interactive workflow guide. Use when user is unsure what to do next, needs help navigating AgentOps, or wants to understand available tools.

install

source · Clone the upstream repo

git clone https://github.com/majiayu000/claude-skill-registry

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data/agent-ops-guide" ~/.claude/skills/majiayu000-claude-skill-registry-agent-ops-guide && rm -rf "$T"

manifest: skills/data/agent-ops-guide/SKILL.md

AgentOps Workflow Guide

Purpose

Help users navigate the AgentOps workflow by asking diagnostic questions and recommending the appropriate next step.

When to Use

User says "help", "what should I do", "where do I start"
User seems lost or confused about workflow
User wants to understand available tools
First time using AgentOps in a project

Diagnostic Procedure

Step 1: Assess State Files

First, silently check which state files exist:

□ .agent/constitution.md  → Project setup complete?
□ .agent/baseline.md      → Baseline captured?
□ .agent/focus.md         → Has session context?
□ .agent/issues/          → Has defined issues?
□ .agent/memory.md        → Has learned conventions?

Step 2: Ask Situational Question

Ask ONE question to understand user's intent:

"What brings you here today?"

A) Starting a new project or first time here B) Returning to continue previous work C) Have a specific task or feature to implement D) Something's broken and need to fix it E) Want to explore or understand the codebase F) Need to review code quality G) Wrapping up work, ready to commit H) Want to create a new Python project I) Need to review/audit an API

Step 3: Recommend Based on State + Intent

Intent	Missing Constitution	Missing Baseline	Has Tasks	Recommendation
A (new)	✗	—	—	`/agent-init` then `/agent-constitution`
A (new)	✓	✗	—	`/agent-baseline`
B (resume)	—	—	—	Read focus.md, summarize status
C (task)	✗	—	—	`/agent-constitution` first
C (task)	✓	✗	—	`/agent-baseline` first
C (task)	✓	✓	✗	`/agent-task` to define task
C (task)	✓	✓	✓	`/agent-plan` for next task
D (broken)	—	—	—	`/agent-debug` then `/agent-recover`
E (explore)	—	—	—	`/agent-map` or `agent-ops-critical-review`
F (review)	—	—	—	`/agent-review` or `/agent-validation`
G (finish)	—	✗	—	`/agent-baseline` then `/agent-review`
G (finish)	—	✓	—	`/agent-validation` then `/agent-review`
H (python)	—	—	—	`/agent-create-python-project`
I (api)	—	—	—	`/agent-api-review`

Step 4: Provide Clear Next Step

Format your recommendation as:

## Your Situation
[One sentence summary of what I detected]

## Recommended Next Step
**Run:** `/agent-[command]`
**Why:** [Brief reason]

## After That
[What comes next in the workflow]

Decision Trees

"I'm new here"

Has .agent/ folder?
├─ NO → /agent-init
└─ YES
   Has constitution.md with CONFIRMED commands?
   ├─ NO → /agent-constitution
   └─ YES
      Has baseline.md?
      ├─ NO → /agent-baseline
      └─ YES → Ready! Ask what they want to build

"I have a task"

Task well-defined (clear acceptance criteria)?
├─ NO → /agent-task (refine it)
└─ YES
   Have baseline?
   ├─ NO → /agent-baseline
   └─ YES
      Have approved plan?
      ├─ NO → /agent-plan
      └─ YES → /agent-implement

"Something's broken"

What's broken?
├─ Build fails → /agent-debug then /agent-recover
├─ Tests fail → /agent-debug (compare to baseline)
├─ Agent stuck → Read focus.md, identify blocking issue
├─ Git issues → agent-ops-git skill
└─ Unknown cause → /agent-debug (systematic isolation)

"I want to create a Python project"

Have requirements/discussion?
├─ YES → /agent-create-python-project (with input)
└─ NO → /agent-create-python-project (will interview)

"I need to review an API"

Has OpenAPI spec?
├─ YES → /agent-api-review
└─ NO 
   Has API endpoints?
   ├─ YES → /agent-api-review (will identify spec gaps)
   └─ NO → Not an API project, use /agent-review

"I'm done"

Changes validated?
├─ NO → /agent-validation
└─ YES
   Critical review done?
   ├─ NO → /agent-review
   └─ YES
      Retrospective done?
      ├─ NO → /agent-retrospective
      └─ YES → Ready to commit (with confirmation)

Quick Reference Card

Present this when user asks for overview:

┌───────────────────────────────────────────────────────────────┐
│                     AgentOps Workflow                         │
├───────────────────────────────────────────────────────────────┤
│  SETUP          │  WORK             │  FINISH                 │
│─────────────────│───────────────────│─────────────────────────│
│  /agent-init    │  /agent-task      │  /agent-validation      │
│  /agent-const   │  /agent-plan      │  /agent-review          │
│  /agent-base    │  /agent-impl      │  /agent-retrospective   │
├───────────────────────────────────────────────────────────────┤
│  UTILITIES                                                    │
│───────────────────────────────────────────────────────────────│
│  /agent-help        Interactive guide (you are here)          │
│  /agent-map         Understand codebase                       │
│  /agent-testing     Test strategy                             │
│  /agent-spec        Manage requirements                       │
│  /agent-report      View issues and status                    │
│  /agent-version     Versioning and changelog                  │
├───────────────────────────────────────────────────────────────┤
│  SPECIALIZED                                                  │
│───────────────────────────────────────────────────────────────│
│  /agent-debug                Systematic debugging             │
│  /agent-api-review           API contract & behavior audit    │
│  /agent-create-python-project   Scaffold Python project       │
└───────────────────────────────────────────────────────────────┘

Anti-Patterns

❌ Overwhelming user with all options at once
❌ Recommending steps when prerequisites aren't met
❌ Skipping constitution/baseline for "quick" tasks
❌ Assuming user knows the workflow

Output

After guiding user, update

.agent/focus.md

## Doing now
- Guided user to [recommended step]
- Reason: [why this step]