Automation on Corey Daley

Before the First Commit: What Multi-Agent Sprint Planning Actually Catches

Sat, 11 Apr 2026 18:05:00 -0400

What does a multi-agent sprint planning workflow actually produce? Not just a cleaner document — it finds real bugs in a plan before implementation begins. When the /sprint-plan command ran against a “simple” Go REST API project, the security review phase returned three critical findings: a logical contradiction that made the stated auth behavior impossible, a schema constraint that would silently break token revocation, and a SQLite pragma applied to only one connection in a pool.

The post walks through the entire planning session for org-api — from seed prompt to approved sprint document — showing what each phase of the review pipeline produced and what changed as a result. The security findings came from reading the plan carefully, not from running any code. That’s the point.

What step in your planning process is explicitly there to prove the plan wrong before implementation begins?

From Config Hub to Competing Voices: How agent-config Became My AI Collaboration Stack

Sat, 11 Apr 2026 14:50:00 -0400

I started agent-config as a shared configuration hub: one repository to rule Claude, Codex, Copilot, and Gemini. That lasted about two iterations before the cracks showed. Forcing every AI agent to share the same configuration format was the wrong abstraction — different tools, different philosophies, different file formats. The solution wasn’t more uniformity. It was a different model of collaboration entirely.

Today agent-config is Claude-specific, but Codex is still central to how I work. The difference: Codex is no longer a configuration target. It’s a competitive collaborator. Sprint plans, blog posts, security audits — every significant output runs through a workflow where Claude and Codex produce independent drafts, critique each other’s work, and force synthesis from the tension. Two AI voices with different instincts produce better output than either would alone — just like a team of people with different backgrounds does.

Is your multi-agent workflow built for sharing configuration, or for generating the productive disagreement that makes output actually better?

Polyphon at v0.8.0: The End of the Prototype Phase

Mon, 23 Mar 2026 19:55:00 -0400

When I shipped Polyphon v0.1.0-alpha.2, the pitch was simple: put multiple AI voices in one conversation and let them respond to each other. That was useful. But early usefulness and long-term trust are not the same thing.

v0.8.0 is the release where Polyphon crosses that line. The features that made the difference weren’t the ones I planned at launch. Voices can now interact with actual files, with per-voice sandboxing and explicit permission categories. Conversation history is encrypted with SQLCipher whole-database AES-256, with optional password protection. FTS5 search turns the archive into working memory you can actually retrieve from. These aren’t incremental improvements — they’re the features that decide whether a tool stays an interesting experiment or earns a place near real projects.

What actually makes you trust an AI tool with real work: capability, privacy, or memory?

Polyphon's MCP Server Makes Your AI Ensemble a Native Agent Tool

Mon, 23 Mar 2026 19:35:00 -0400

Polyphon just shipped MCP server support, and it reframes what the tool actually is. You’ve been opening it as a UI. Now your coding agent — Claude Code, Cursor, Codex CLI, GitHub Copilot — can call into it directly and broadcast a prompt across your entire Claude + GPT-4o + Gemini ensemble in a single tool call.

The practical win is not just convenience. It is that you can package disagreement: three independent models, each with different priors, applied to the same question at once. Five tools, two CLI flags, one Settings toggle. The SessionEventSink abstraction is why this works without duplicating orchestration logic — desktop UI and headless agent calls run on the same engine.

Where in your workflow would paying for disagreement actually change a decision, rather than just produce a louder answer?

I Built a Tool So AI Models Could Talk to Each Other

Mon, 16 Mar 2026 12:30:00 -0400

Every AI power user I know runs the same manual workaround: ask Claude, ask GPT, copy the interesting parts of each into the other, then try to synthesize what you learned. The models are good. The coordination is not.

I just shipped Polyphon v0.1.0-alpha.2 — a free, local-first desktop app that puts multiple AI voices in the same conversation so they can actually respond to each other. You’re the conductor. They’re the ensemble. Save a group of voices as a composition and reuse it whenever you need that ensemble again.

What should a multi-agent conversation feel like when you’re not building a pipeline — when you just want to think out loud with several models at once?

The Blog That Builds Itself: AI Automation Behind the Scenes

Tue, 10 Mar 2026 15:40:00 -0400

What happens when you stop treating AI as a writing assistant and start treating it as a co-publisher? This blog has accumulated a full automation stack: a /create-blog-post command that pits Claude and Codex against each other in a competitive draft workflow, a Python script that generates hero images using DALL-E 3 and Claude vision, pre-commit hooks that block commits with unoptimized images, and a Bash script that handles WebP conversion and thumbnail generation automatically.

The design principle behind all of it: treat each stage as a contract, not a prompt. The meta-detail: this post was written by the same pipeline it describes.

What would your content workflow look like if you designed it the same way you’d design a software system — and which parts would you never automate at all?

Why I Left Notion and Built My Own AI Agent Plugin for Obsidian

Mon, 09 Mar 2026 11:45:00 -0400

I was already paying for Claude, Codex, and several other AI services when Notion started pushing its own AI add-on. The problem wasn’t just the price — it was that Notion AI couldn’t talk to any of the tools I already had, and my workflow had become a copy-paste treadmill between my notes and my agents. So I switched to Obsidian and hit the same wall: no native way to use your own AI.

I built the Obsidian AI Agent Sidebar plugin — an open source tool that brings Claude Code, OpenAI Codex, Google Gemini, GitHub Copilot, and any OpenAI-compatible server directly into your Obsidian sidebar, with real vault read/write access. Developed using agentic engineering and the dark factory method, it’s the integration I needed and couldn’t find.

What would you build if the tool you needed simply didn’t exist yet?

When Your First Version Fails: Iterating on agent-config with AI

Sat, 07 Mar 2026 16:00:00 -0500

I built agent-config v1 to centralize AI agent configurations across Claude, Codex, Copilot, and Gemini — and it failed. Not dramatically, but fundamentally: I tried to force every agent to follow the same rules in the same format, because that’s what my human instincts said made sense. The problem is those agents have completely different requirements. Gemini needs TOML. The others use Markdown. You can’t just symlink your way to consistency.

v2 fixes this by letting AI handle the translation — automated merging of global and per-agent configs, format conversion per tool, and intelligent symlink setup.

The real lesson isn’t about config management, though. It’s about failing fast, iterating faster with AI than you ever could alone, and trusting the tools to solve problems your instincts would have you paper over. Are you letting your human instincts slow down your AI iteration cycles?

Attractor: When Chaos Theory Meets AI Pipeline Orchestration

Fri, 06 Mar 2026 12:00:00 -0500

Most AI automation breaks for the same reason: the workflow is hidden inside scripts, prompts, and tribal knowledge. Attractor takes a different path — one borrowed from chaos theory. In dynamical systems, an attractor is a state a system naturally converges toward even through turbulence.

Apply that to AI workflows, and you get directed graphs that pull multi-LLM execution through branching, retries, human review gates, and failure recovery toward a defined goal. The entire codebase is AI-generated, which makes it a working proof-of-concept of the Software Factory philosophy.

If we can declaratively orchestrate AI agents today, what should we still insist on owning as humans?

Centralizing AI Agent Configurations with the agent-config Repository

Thu, 19 Feb 2026 20:26:46 -0500

If you’re juggling Claude Code, Codex, and GitHub Copilot, you know the pain of keeping each one’s configuration files in sync. My agent-config repo solves that with a single source of truth: agent instructions, reusable skills, custom commands, subagents, and prompts all live in one place, and a single make symlinks command wires them up across every tool.

Each directory is purpose-built—skills for reusable capabilities, commands for CLI tools, subagents for delegation, prompts for task-specific guidance. The setup even backs up any files it would overwrite, so you never lose existing config.

Whether you’re just starting to extend your AI tools or already deep into custom workflows, having everything version-controlled and centralized is a game changer. Are you managing your AI agent configurations in a single repository, or do you keep them scattered across tools?

Automate Your Blog with Notion and AI: A Self-Demonstrating Workflow

Tue, 10 Feb 2026 21:41:59 -0500

This post you’re reading right now? It was created by an AI reading a to-do item from my Notion database. That’s the power of combining Notion with AI assistants. The problem every blogger faces: brilliant ideas die in the gap between inspiration and execution.

My solution: a Notion to-do list where I capture ideas, and AI assistants (Claude Code and ChatGPT) read from it via Model Context Protocol, generate complete posts, publish them to my Hugo blog, and mark the to-dos complete. It’s self-demonstrating—this very post was created that way. The workflow transforms content creation from manual drudgery into an automated pipeline.

If you’re drowning in blog ideas but low on execution energy, this might be your answer. Are you using Notion for content management? Have you explored AI integrations for your blog?

Managing Blog Posts with GitHub Copilot

Wed, 04 Feb 2026 16:25:00 -0500

Managing a blog used to mean juggling multiple tools—notes apps for ideas, editors for drafting, task trackers for progress. It was fragmented and exhausting. Then I discovered GitHub Copilot can work directly with GitHub Issues and Projects, creating a seamless workflow.

Here’s how it works: I create issues for blog post ideas, GitHub Copilot reads the issue, generates the complete post with proper frontmatter and content, and automatically closes the issue when I publish. No context switching, no lost ideas, just a smooth pipeline from concept to publication. The best part? Everything lives in one place alongside my code.

If you’re blogging and using GitHub, this workflow is a game-changer. How are you managing your content pipeline? Have you tried integrating your blog workflow with your code repository?