osop（osop 分析）

OSOP — Open Standard Operating Procedures

Universal protocol for defining, validating, risk-assessing, and executing process workflows. Works with any AI coding agent.

Capabilities

When this skill is active, the agent can:

• Create OSOP workflow definitions from natural language
• Validate workflow YAML against the OSOP schema
• Risk Assess workflows for security issues, permission gaps, destructive commands
• Execute workflows with dry-run mode for safety
• Render workflows as Mermaid diagrams
• Optimize workflows using execution history — detect slow steps, failure hotspots
• Convert between OSOP and external formats (GitHub Actions, BPMN, Airflow)
• Report on workflow executions as standalone HTML

Sub-Skills

This pack includes 4 specialized skills:

OSOP Node Types (v1.0)

16 supported node types in 4 categories:

Actors: human, agent, company, department Technical: api, cli, db, git, docker, cicd, infra, mcp Flow Control: system, event, gateway, data

Use subtype for domain specialization (e.g., type: agent, subtype: llm).

Edge Modes

13 modes: sequential, conditional, parallel, loop, event, fallback, error, timeout, compensation, message, dataflow, signal, weighted, spawn.

Quick Start

1. Define a workflow

osop_version: "1.0"
id: "deploy-staging"
name: "Deploy to Staging"
description: "Build, test, and deploy to staging environment."
version: "1.0.0"
tags: [deploy, staging]

nodes:
  - id: "build"
    type: "cli"
    name: "Build Project"
    description: "Run the build command"
  - id: "test"
    type: "cicd"
    subtype: "test"
    name: "Run Tests"
    description: "Execute test suite"
  - id: "deploy"
    type: "infra"
    name: "Deploy to Staging"
    description: "Push to staging environment"
    security:
      risk_level: "medium"
      approval_gate: true

edges:
  - from: "build"
    to: "test"
    mode: "sequential"
  - from: "test"
    to: "deploy"
    mode: "conditional"
    when: "tests.passed == true"

2. Review for risks

Run /osop-review deploy-staging.osop to check for security issues before execution.

3. Execute and log

After running, use /osop-log to record what happened as a structured .osoplog.yaml.

4. Generate report

Run /osop-report to create a standalone HTML report with dark mode and expandable nodes.

Session Logging

After completing multi-step tasks, this skill produces:

• .osop — workflow definition (what should happen)
• .osoplog.yaml — execution record (what actually happened)

These files can be visualized at https://osop-editor.vercel.app or converted to HTML.

Security Metadata

Nodes can declare security.risk_level (low/medium/high/critical), security.permissions, security.secrets, and approval_gate for risk analysis.

Self-Optimization Loop

1. Execute workflow → produce .osoplog
2. Aggregate stats from past runs
3. Run /osop-optimize → get improvement suggestions
4. Apply approved changes → execute improved workflow
5. Repeat — the SOP gets better each iteration

Links

• Spec: https://github.com/Archie0125/osop-spec
• Visual Editor: https://osop-editor.vercel.app
• Examples: https://github.com/Archie0125/osop-examples
• Website: https://osop.ai