Carabiner eliminates the protocol ceremony of writing Claude Code hooks. You write a typed function. Carabiner handles stdin, stdout, JSON formatting, exit codes, error routing, timeouts, and logging. The logic is yours. The plumbing is ours.
Zero external dependencies. Built on Bun.
Claude Code quietly shipped something I've been waiting for: agents that learn between sessions. Not "memory" in the chatbot sense — actual persistent knowledge stores that agents write to disk and carry forward into future work.
There are two flavors. Auto memory is for the main Claude session itself. Subagent memory is for individual subagents you define. Both use the same architecture under the hood, and both are worth understanding if you're building anything with agents that needs to get smarter over time.
Auto memory is a persistent directory where Claude records patterns, insights, and project-specific knowledge as it works. This is different from CLAUDE.md files (which are instructions you write for Claude). Auto memory is notes Claude writes for itself.
GitButler is the best local stack experience that exists. Full stop. Parallel branches, auto-restacking mid-stack edits, absorb, the new CLI — all of it solves a problem that Git itself has ignored for 20 years. I’ve been using it daily and the workflow just clicks in a way that Graphite’s local tooling never quite did.
But here’s the thing: the moment I push a stack to GitHub, I’m on my own.
PRs land on GitHub looking like unrelated branches unless you squint at base refs. When the bottom PR merges, the next one’s diff bloats with already-merged changes (or worse, targets a deleted branch). Mid-stack review feedback means I need to rebase locally, force-push, and hope GitHub’s UI catches up. And if I’ve got an agent working on a fix mid-stack? I’m the router — copying context, rebasing its work, making sure the rest of the stack stays coherent.
Research notes for the Outfitter Stack. Synthesized from codebase analysis, MCP spec, CLI guidelines, and logging library documentation.
Output is the product — what the user asked for. Logging is the process — how the program ran.
| Concern | Stream | Audience | Example |
|---|
obc-cli is a command-line interface for managing Obsidian vaults. Its defining design principle is filesystem-first: the vast majority of operations work directly on vault files without Obsidian running. The Obsidian Local REST API plugin is only required for two capabilities: Dataview DQL queries and Obsidian command execution.
This reverses the typical dependency model. Most Obsidian tooling assumes the app is running — obc assumes it isn't, and gracefully adds REST API features when available.
TRADITIONAL: CLI → REST API (required) → Obsidian → Files
A tiny, append-only micro-log intended to act as a “continuity spine” across fragmented contexts (Slack threads/DMs/channels, multiple sessions, background runs).
Core idea:
One JSON object per line.
Design doc for enriching sources with full content via targeted crawling.
BLZ (pronounced "blaze") is a local-first search cache for llms.txt documentation. It keeps documentation local, searches it in milliseconds (P50 ≈ 6ms), and returns grounded spans with exact line citations.
I'm exploring integrating agentation into Navigator, a browser automation tool for AI agents. Navigator has a "paired mode" where agents work alongside users in their browser — agentation's annotation UX would be a perfect fit for letting users visually mark elements for agents to interact with.
Agentation currently outputs annotations via navigator.clipboard.writeText(). For integration into other tools, this requires intercepting the clipboard API, which is: