Perception on Start AI Tools - Presented by Intent Solutions

git-with-intent Competitive Strategy and Perception Real Ingestion Pipeline

Fri, 13 Feb 2026 10:00:00 -0600

Two commits across two repos. A light day by commit count, but both commits landed meaningful capability that had been blocking next steps. git-with-intent got its competitive positioning documented — which unblocked the pricing page design and the README rewrite that followed later in the month. Perception closed the loop between the dashboard button and real article storage, which meant the product was finally demo-able without apologies about mock data.

IRSB Four Releases in One Day, Perception Topic Watchlist, and Hustle Session Cookies

Sun, 08 Feb 2026 10:00:00 -0600

Three repos, seventeen commits. The IRSB monorepo pushed through four releases in a single day — clearing a backlog of fixes that had been accumulating across the solver, dashboard, SDK, and protocol packages. Perception moved from static topic display to interactive CRUD with real Firestore persistence. Hustle fixed the auth flow that had been silently expiring users after one hour and stabilized two more flaky E2E tests.

IRSB: v0.2.0 Through v1.2.0

Ten commits, four releases. The IRSB monorepo had accumulated fixes across multiple packages — TypeScript strict mode violations, ERC-8004 adapter badge rendering, and a Lit network key derivation bug that only surfaced on Sepolia. Instead of batching these into a single large release, I cut four separate versions to keep the changelog readable and bisectable.

IRSB Agent Passkey Scaffold on Cloud Run and Perception v0.6.0 Trending

Fri, 06 Feb 2026 10:00:00 -0600

Two projects, twelve commits total, and each shipped something tangible. IRSB got its first agent authentication scaffold. Perception tagged a release. Neither project had more than six commits, but both moved from “in progress” to “deployed.”

IRSB: Agent Passkey on Cloud Run

Six commits building the agent-passkey service from scratch and deploying it to Cloud Run.

The Problem

IRSB’s agents need to authenticate against the protocol. Traditional API keys don’t work well for autonomous agents — keys get leaked in logs, rotated keys break running agents that cached the old key, and there’s no way to scope permissions per agent instance without building a custom authorization layer on top.

IRSB v0.1.0 Milestone, Perception YAML Feeds, and HN Expansion

Thu, 05 Feb 2026 10:00:00 -0600

Two projects dominated February 5th. IRSB hit its first tagged release. Perception expanded its feed coverage by 72%.

IRSB Monorepo: v0.1.0

Fourteen commits pushing the monorepo to its first milestone release. The version number is modest — v0.1.0, not v1.0.0 — because the protocol’s smart contracts haven’t been audited by a third party yet. But the infrastructure around the contracts is now release-quality.

Phase 3-7 AARs

After-Action Reviews for each completed development phase. These aren’t retrospective essays — they’re structured documents covering what was planned, what actually happened, what went wrong, and what changes to apply in the next phase.

New Year's Eve: A Go CLI From Scratch, a Security Fix, and Two Releases

Wed, 31 Dec 2025 10:00:00 -0600

Last day of 2025. Five projects touched. Twenty-five commits. One project built from nothing, one patched a security hole that should never have existed, and two shipped point releases to close out the year clean.

create-agent-skill: New Go CLI

Nine commits. Brand new project. A Go CLI tool for validating and scaffolding Claude Code agent skills. The problem it solves: skill definitions are YAML files with a specific schema, and getting the schema wrong means the skill silently fails to load or behaves unexpectedly at runtime.

Dream Gym Sprints, Email AI Stack, and 128 RSS Feeds

Mon, 29 Dec 2025 10:00:00 -0600

Four projects. Thirty-eight commits. Two of them are building real product features, one is growing into an intelligence platform, and one just needed a Dockerfile fix.

Dream Gym: Sprints 1 Through 3

Hustle’s Dream Gym module went from concept to working product in three sprints and fourteen commits.

Sprint 1: Workout Logging. The core data model is straightforward — exercises, sets, reps, weight, duration. But the UX question was whether to optimize for speed or detail. Speed won.