10 AI Agents · 2 Pipelines · Zero Manual DevRel
Your DevRel Pipeline,
Running by Friday
Replace your DevRel, content, and sales teams with an autonomous agent system that triages issues, writes tutorials, records videos, gathers competitive intel, and runs outreach — all coordinated through shared context.
The DevTools Advocate Agent is a production-grade multi-agent AI system built on Claude Agent SDK and Model Context Protocol (MCP). It comprises 10 autonomous agents across 2 pipelines (DevRel + Sales), coordinating via 14 MCP-registered tools, grounded in 18 curated knowledge base documents, and validated by 337 automated test cases.
10
Autonomous Agents
2
Pipelines
14
MCP Tools
18
Knowledge Base Docs
337
Test Cases
Architecture
Hub-and-Spoke
Orchestration
Atlas coordinates a weekly cycle across two pipelines — DevRel and Sales. Upstream outputs flow downstream: triage and social signals inform synthesis, synthesis informs experiments and competitive intel, which inform content, video, docs, outreach, and campaigns.
Atlas
Orchestrator · Delegation · OKR Tracking
Sage
Community Manager
Echo
Social Listener
Iris
Feedback Synthesizer
Nova
Growth Strategist
Rex
Competitive Intel
Kai
Content Creator
Vox
Video Producer
Dex
Doc Generator
Pax
Sales Enablement
Mox
Campaign Marketing
DevRel Pipeline
Sales Pipeline
The Agents
Ten Specialists,
Two Pipelines
Each agent owns a single domain of developer advocacy, with typed inputs/outputs and access to specific tool sets.
Atlas
Orchestrator
Coordinates the weekly cycle, delegates tasks with retry + exponential backoff, brokers cross-agent context, and compiles OKR progress.
delegationretrycontext brokerOKR tracking
Sage
Community Manager
Triages GitHub issues with priority scoring, sentiment analysis, and churn detection. Flags at-risk contributors and identifies community champions.
GitHub APIsentimentpriority scoring14 tools
Iris
Feedback Synthesizer
Extracts recurring themes from upstream signals using LLM analysis. Ranks pain points by composite score and maps them to developer journey stages.
theme extractionjourney mappingpain point ranking12 tools
Nova
Growth Strategist
Designs A/B experiments with statistical rigor — power analysis, sample size calculations, guardrail metrics. Identifies funnel drop-off points.
scipyBayesian evalfunnel analysis14 tools
Echo
Social Listener
Scans Reddit, Hacker News, and Twitter for product mentions. Flags engagement opportunities, reputation risks, and trending discussions.
FirecrawlBrave searchsentimentrisk alerts
Kai
Content Creator
Writes technical tutorials grounded in the knowledge base and informed by upstream pain points. Validates every code block via ast.parse(), delimiter balancing, and json.loads().
knowledge basecode validationGitMCPcontent gen
Vox
Video Producer
Produces screen-recorded tutorials with TTS narration and FFmpeg overlays. Turns written tutorials into step-by-step video walkthroughs.
PlaywrightFFmpegOpenAI TTSoverlays
Dex
Documentation Generator
Parses Python source with ast.parse() and JS/TS with heuristic regex. Generates architecture overviews, API references, and module guides from actual code.
AST parsingarchitecture docsAPI referencemodule guides
Rex
Competitive Intelligence
Competitor discovery, threat assessment, opportunity mapping with market positioning analysis.
competitor trackingthreat assessmentmarket analysisopportunity mapping
Pax
Sales Enablement
Personalized outreach emails, battle cards, nurture sequences, and objection handling playbooks.
outreach emailsbattle cardsnurture sequencesobjection handling
Mox
Campaign Marketing
Blog posts, landing pages, social batches, campaign briefs, and press releases.
blog postslanding pagessocial batchespress releases
Weekly Cycle
Monday to Friday,
Fully Autonomous
Atlas runs a weekly pipeline where each day's output feeds the next. By Friday, every deliverable is grounded in real community signals.
Key Features
Built for Production
Not a prototype. Every pattern is designed for reliability, observability, and portability.
Cross-Agent Context Sharing
Insights cascade between agents via SharedContext. Iris's pain points inform Kai's tutorials and Nova's experiment hypotheses.
Retry with Exponential Backoff
Failed delegations retry up to 2x with jittered delays. Graceful degradation when LLM resources are unavailable.
Typed Async API Client
Full PostHog API v2 coverage with dataclass models. httpx async client with proper error handling throughout.
Statistical Rigor
Nova runs scipy-based power analysis, Bayesian experiment evaluation, and frequentist significance testing.
Code Validation in Content
Kai validates every code block: Python via ast.parse(), JavaScript via delimiter balancing, JSON via json.loads(), HTML via tag checking, SQL via keyword analysis.
Official Docs Grounding
Content agents fetch official documentation from GitMCP and cross-check output against it. No hallucinated API signatures.
AST-Based Documentation
Dex parses Python source with ast.parse() and JS/TS with heuristic regex to extract classes, functions, signatures, and docstrings from actual code.
Pluggable Knowledge Base
18 curated docs covering SDKs, APIs, products, and platform. Swap the knowledge base to retarget at any product.
MCP Tool Protocol
14 tools registered via Model Context Protocol, compatible with Claude Desktop, Cursor, Windsurf, and any MCP client.
Video Production Pipeline
Vox produces screen-recorded tutorials with Playwright, assembles with FFmpeg overlays, and narrates with OpenAI TTS — all automated.
Dual Pipeline Architecture
DevRel and Sales pipelines run in parallel, sharing context. Community insights feed competitive intelligence, which feeds outreach.
Competitive Intelligence
Rex discovers competitors from community signals and web search, assesses threats, and maps opportunities for positioning.
Sample Output
Every Deliverable Is
Generated by the System
These artifacts were produced by a live pipeline run against PostHog's open-source repository and real community data.
KaiTechnical Tutorial — Feature Flag-Driven Onboarding with PostHog + Next.js
→SageWeekly Community Triage Report — W10 2026
→EchoSocial Listening Report — 63 Mentions Across 4 Platforms
→IrisDeveloper Feedback Synthesis — Q1 2026 Themes
→NovaA/B Experiment Pre-Registration with Power Analysis
→VoxVideo Script & Production Plan — Feature Flags Tutorial
→DexArchitecture Overview — Agent System API Reference
→AtlasWeekly Pipeline Summary & OKR Progress — W10 2026
→RexCompetitive Intelligence Brief — Weekly Threat Assessment
→PaxOutreach Sequences & Battle Cards — Sales Enablement Pack
→MoxCampaign Content — Blog Posts, Landing Pages, Social Batches
→Tech Stack
What It's Built With
Agent Execution
Claude Agent SDK
Tool Protocol
Model Context Protocol (MCP)
Language
Python 3.12+ async
LLM Backbone
Claude Sonnet 4.6
HTTP Client
httpx (async)
Statistics
scipy + Bayesian eval
Video
Playwright + FFmpeg + OpenAI TTS
Testing
pytest + respx (337 tests)
Search
Brave + Firecrawl
Retargetable
Point It at Any
DevTools Product
The system is product-agnostic. I retarget it to your product during onboarding — here's how.
01
Swap Knowledge Base
Load your SDK docs, API reference, and product documentation into the agent's knowledge base.
02
Connect Your APIs
Point the API client at your product's endpoints and configure data models.
03
Target Your Repo
Connect the agents to your GitHub repository for community triage and social listening.
04
Tune Agent Prompts
Customize agent personas, tone, and product-specific context to match your brand.
05
Launch the Cycle
Start the weekly pipeline and receive autonomous deliverables every Friday.
Pricing
Choose Your Plan
Every plan includes the full 10-agent pipeline, 11 weekly deliverables, and the "Run It First" guarantee.
Full-Time DevRel Hire
$12,000+/mo
Salary + benefits + ramp time
DevRel Agency
$15,000+/mo
Retainer + slow iteration
Fractional CMO
$8,000+/mo
Strategy only, no execution
AI Agent System
from $5,000/mo
Full pipeline + execution + guarantee
Pilot
$5,000/mo
3-month commitment
- Full 10-agent pipeline on your repo
- 11 deliverables per week
- Weekly Slack report with key findings
- 1 strategy call per month
"Run It First" guarantee included
Bonus: Free "Community Health Audit"
Start Pilot →Growth
$7,500/mo
Everything in Pilot plus deeper integration
- Custom agent tuning for your product voice
- Content calendar aligned with your roadmap
- Growth experiments with statistical rigor
- Bi-weekly strategy calls
- Priority Slack support
"Run It First" guarantee included
Bonus: LinkedIn content repurposing
Start Growth →Enterprise
$10,000/mo
For Series A+ companies
- Custom agent development for your workflows
- Quarterly roadmap reviews
- Dedicated async channel
- White-label deliverables
"Run It First" guarantee included
Bonus: Competitor monitoring
Start Enterprise →The "Run It First" Guarantee
I deploy the full agent system on your repo and run it for 2 weeks. If the output isn't good enough to use — if you wouldn't put your name on the tutorials, triage reports, and experiment designs — you pay nothing for the first month.
No contracts. No fine print. If the agents don't deliver, I eat the cost.
Why I can offer this: the system has 337 tests, validates every code block, and grounds all content in official documentation. I've run it against PostHog's 40K+ star repo. I'm confident in what it produces.
How It Works
What Happens After
You Say Yes
You don't configure anything. Here's the full onboarding timeline.
Day 0
Book a Call
30-minute call. I ask about your product, repo, and what's broken in your DevRel today.
Day 1
Free Audit
I run 3 agents on your repo and send you the report. No commitment required.
Day 1–2
Load Your Context
I load your docs, API reference, and product context into the knowledge base.
Day 3
Connect & Configure
Agents connect to your GitHub repo. Social listening configured for your product.
Day 4
Tune & Test
Agent prompts tuned to match your product voice and terminology. End-to-end test run.
Day 5
First Deliverables
First pipeline run. 8 deliverables land in your inbox. Zero effort on your side.
FAQ
Common Questions
What is the DevTools Advocate Agent?+
The DevTools Advocate Agent is a production-grade multi-agent AI system built on Claude Agent SDK and Model Context Protocol (MCP). It consists of ten specialized agents across two pipelines — DevRel (Atlas, Sage, Echo, Iris, Nova, Kai, Vox, Dex) and Sales (Rex, Pax, Mox) — that coordinate a fully autonomous weekly developer advocacy and sales enablement pipeline.
How does this compare to hiring a human developer advocate?+
The agent system handles the repetitive, high-volume tasks of developer advocacy — scanning 60+ social mentions, triaging 39+ GitHub issues, and producing tutorials with validated code blocks every week in under 19 minutes of total execution time. A human DevRel lead still sets strategy, reviews output, and handles relationship-building. The agent handles the 80% that's systematic; humans handle the 20% that requires judgment.
What technology stack does the agent system use?+
Built on Claude Agent SDK with Model Context Protocol (MCP) for tool integration. Python 3.12+ async, Claude Sonnet 4.6 as the LLM backbone, httpx for async HTTP, scipy for statistical analysis, Playwright + FFmpeg + OpenAI TTS for automated video production, and pytest with 337 test cases for validation.
Can this be retargeted to my product?+
Yes. The system is product-agnostic. Retargeting involves five steps: swapping the knowledge base with your SDK docs and API reference, connecting your product APIs, pointing agents at your GitHub repository, tuning agent prompts for your brand voice, and launching the weekly cycle. The entire retargeting process is handled during onboarding.
What deliverables does the system produce each week?+
Each weekly pipeline produces: a community triage report with churn risk flags, a social listening report covering Reddit/HN/Twitter/Dev.to, a developer feedback synthesis with ranked pain points, A/B experiment pre-registrations with power analysis, a technical tutorial with AST-validated code blocks, a video production plan with scene scripts, architecture documentation parsed from source code, competitive intelligence briefs, outreach email sequences with battle cards, multi-channel campaign content (blog posts, landing pages, social batches), and an OKR progress summary.
Ready to Stop Doing DevRel at Midnight?
Book a 30-minute call. I'll ask about your product, your repo, and what's broken — then I'll run a free audit so you can see the output before committing.