Competitor Analysis
Orchestrate 7 research agents to produce a comprehensive competitor teardown. Phase 1 runs 4 parallel research agents (Product, Marketing, UX, Technical). Phase 2 runs 3 sequential synthesis agents (MVP Spec, GTM Strategy, Competitive Landscape) that build on Phase 1 findings.
Workflow
1. Parse Input
Target: $ARGUMENTS
If the target above is non-empty, use it immediately — do NOT ask the user to confirm or re-provide it. Parse it as follows:
- URL (starts with
http): use as-is for WebFetch, extract company name from domain - Company name (no URL): construct likely URLs (
https://{name}.com,https://www.{name}.com)
If the target above is empty, ask the user what competitor to analyze and wait for their response.
Store the parsed values:
COMPANY_NAME: Human-readable name (e.g., "Linear")PRIMARY_URL: Main product URL (e.g., "https://linear.app")
IMPORTANT: When a target is provided, begin Phase 2 immediately after parsing. Do not pause for user input.
2. Phase 1 — Parallel Research Agents
Spawn 4 agents in parallel using the Task tool. Each agent is general-purpose (needs WebSearch + WebFetch). Run all 4 with run_in_background: true for maximum parallelism.
Read REFERENCE.md first to get the detailed research checklists and output templates for each agent.
Agent 1: Product Overview
subagent_type: general-purpose
run_in_background: truePrompt:
You are a product research analyst. Research {COMPANY_NAME} ({PRIMARY_URL}) and produce
a comprehensive product overview.
Follow the "Product Overview Agent" template in the reference below. Use WebSearch and
WebFetch to gather information. Cite sources for every claim.
{paste Product Overview section from REFERENCE.md}Agent 2: Marketing Analysis
subagent_type: general-purpose
run_in_background: truePrompt:
You are a marketing strategist. Research {COMPANY_NAME}'s marketing and positioning.
Follow the "Marketing Analysis Agent" template in the reference below. Use WebSearch and
WebFetch to analyze their marketing channels, messaging, and content strategy.
{paste Marketing Analysis section from REFERENCE.md}Agent 3: UX Analysis
subagent_type: general-purpose
run_in_background: truePrompt:
You are a UX researcher. Analyze the user experience of {COMPANY_NAME} ({PRIMARY_URL}).
Follow the "UX Analysis Agent" template in the reference below. Use WebFetch to walk
through their signup flow, onboarding, and core product experience.
{paste UX Analysis section from REFERENCE.md}Agent 4: Technical Stack
subagent_type: general-purpose
run_in_background: truePrompt:
You are a technical researcher. Investigate the technology stack behind {COMPANY_NAME}.
Follow the "Technical Stack Agent" template in the reference below. Use WebSearch and
WebFetch to analyze their tech choices, APIs, architecture signals, and engineering culture.
{paste Technical Stack section from REFERENCE.md}3. Collect Phase 1 Results
Wait for all 4 background agents to complete. Read their output files to collect results.
Compile a Phase 1 Summary containing the key findings from each agent. This summary feeds into Phase 2 agents.
4. Phase 2 — Sequential Synthesis Agents
Phase 2 agents run sequentially because each builds on prior results. These are NOT background agents — wait for each to complete before spawning the next.
Read REFERENCE.md for detailed templates.
Agent 5: MVP Specification
subagent_type: general-purposePrompt:
You are a product strategist. Based on the competitor research below, define an MVP
specification for a product that competes with {COMPANY_NAME}.
## Phase 1 Research Findings
{paste compiled Phase 1 findings}
Follow the "MVP Specification Agent" template in the reference below.
{paste MVP Specification section from REFERENCE.md}Agent 6: Go-to-Market Strategy
subagent_type: general-purposePrompt:
You are a go-to-market strategist. Based on the competitor research and MVP spec below,
design a go-to-market strategy for competing with {COMPANY_NAME}.
## Phase 1 Research Findings
{paste compiled Phase 1 findings}
## MVP Specification
{paste Agent 5 output}
Follow the "Go-to-Market Strategy Agent" template in the reference below.
{paste GTM Strategy section from REFERENCE.md}Agent 7: Competitive Landscape
subagent_type: general-purposePrompt:
You are a market analyst. Based on all prior research, map the competitive landscape
around {COMPANY_NAME} and identify differentiation opportunities.
## Phase 1 Research Findings
{paste compiled Phase 1 findings}
## MVP Specification
{paste Agent 5 output}
## Go-to-Market Strategy
{paste Agent 6 output}
Follow the "Competitive Landscape Agent" template in the reference below.
{paste Competitive Landscape section from REFERENCE.md}5. Final Synthesis
Combine all 7 agent outputs into a single report. Present to the user with this structure:
# Competitor Teardown: {COMPANY_NAME}
## Executive Summary
[3-5 bullet points: what they do, how they win, where they're vulnerable]
## Table of Contents
1. Product Overview
2. Marketing Analysis
3. UX Analysis
4. Technical Stack
5. MVP Specification
6. Go-to-Market Strategy
7. Competitive Landscape
---
[Agent 1 output — Product Overview]
---
[Agent 2 output — Marketing Analysis]
---
[Agent 3 output — UX Analysis]
---
[Agent 4 output — Technical Stack]
---
[Agent 5 output — MVP Specification]
---
[Agent 6 output — Go-to-Market Strategy]
---
[Agent 7 output — Competitive Landscape]
---
## Key Takeaways
### Top 3 Opportunities
1. [Biggest gap or underserved segment]
2. [Second opportunity]
3. [Third opportunity]
### Top 3 Risks
1. [Biggest risk in competing]
2. [Second risk]
3. [Third risk]
### Recommended Next Steps
1. [Most important action]
2. [Second action]
3. [Third action]Example Invocations
/competitor-analysis https://linear.app
/competitor-analysis Notion
/competitor-analysis https://www.figma.com
/competitor-analysis VercelAnti-Patterns
- Don't skip Phase 1 before Phase 2: Synthesis agents need research findings to produce useful output. Never run Phase 2 agents without passing them Phase 1 results.
- Don't use Explore agents: Sub-agents need WebSearch and WebFetch for external research. Use
general-purposeonly. - Don't collapse agents: Each agent has a distinct research lens. Combining them loses depth.
- Don't fabricate data: If an agent can't find information (e.g., pricing not public), it should say so explicitly rather than guessing.
- Don't skip citations: Every factual claim must reference a source URL or page.
- Don't run Phase 2 in parallel: Agent 6 needs Agent 5's output, Agent 7 needs both.
Notes
- Total runtime is typically 3-8 minutes depending on the target's web presence.
- Phase 1 agents run in background for parallelism; Phase 2 agents run sequentially.
- If a Phase 1 agent fails or returns thin results, note the gap in the final report rather than blocking Phase 2.
- For private/stealth companies with minimal web presence, agents will produce thinner reports — this is expected.