seo-supervisor

name	seo-supervisor
description	Master Supervisor orchestrator that bridges the SEO and Blog subsystems. Analyzes multi-system prompts, generates roadmaps, coordinates task handoffs via .seo-cache, and executes a lightweight structured review loop with safety caps.
user-invokable	true
argument-hint	<goal> [--max-steps <int>] [--force-fresh]
license	MIT
metadata	{"author":"dotusmanali","version":"1.0.0","category":"seo"}
id	seo-supervisor
domain	ops
invoked_by	["seo-supervisor","seo"]
depends_on	[]

Master Supervisor Orchestrator

Purpose

Acts as the single unified routing layer and coordinator for all SEO and Blog workflows. It bridges the gap between the SEO analysis and Blog creation subsystems, allowing complex multi-step user prompts to run sequentially with cross-system data flow.

When to use

• Anytime the user prompt requires a mix of keyword research/SEO audit and content creation (e.g. "find trends and write a draft").
• To plan and run multi-step sequential tasks interactively across both the SEO and Blog engines.

Command Reference

/seo:run <goal> [--max-steps <int>] [--force-fresh]

Workflow

1. Plan Phase: Parse the user's goal and generate an Execution Roadmap as a visible checklist before executing anything. It details the required skills, target agents, and MCPs/scripts spanning both the SEO and Blog subsystems. Wait for user confirmation unless --yes is set.
2. Cache Check (Fix 3): Before invoking any skill, check .seo-cache/ for existing files matching {domain-or-topic-slug}__{task-type}__{YYYY-MM-DD}.json. If a cache file is less than 7 days old, ask the user if they want to reuse it, unless --force-fresh is set.
3. Execution Loop:
   • Invoke the target subagent or script for the current step (calling either SEO or Blog specialist skills).
   • Cache results in .seo-cache/ to pass context between subsystems (e.g., SEO outputs $\rightarrow$ Blog writer inputs).
4. Lightweight Structured Review (Fix 2):
   • Verify if the expected cache file was written to .seo-cache/ and is valid. Do not re-read full prior outputs.
   • If a step fails, retry it up to a maximum of 2 retries (3 total attempts).
   • If it still fails:
     • If no downstream tasks depend on its output $\rightarrow$ log a Partial Failure and continue.
     • If downstream tasks depend on its output $\rightarrow$ halt and report error.
5. Safety Cap: Capped at max_iterations = 5 total roadmap steps per run. Halt immediately and output partial cache if exceeded.
6. Handoff Phase: Render the final deliverables, highlighting which cached files were reused versus freshly generated.