Writing
What each tool does, what it costs, and what it doesn’t claim.
- A read-only Gmail connector for Claude Code
One OAuth scope, a refresh token in the Keychain, and email the agent treats as data. Also why this site has a privacy policy.
- Seven weeks with a self-improving agent
Nous Research’s Hermes on my Mac: 295 sessions, 105 skills of which it wrote 41, a bot per role, and why it now asks before it runs code.
- Letting an assistant record my money without letting it break the books
Waypoint’s MCP server writes as the signed-in user, through the same database checks as the app, and answers every write with what it changed.
- Two agents, one memory, and why I re-read every write
Hermes and Claude Code share one pool of notes. Mining 148 chat threads into it produced about 400 writes; 361 of them needed fixing when checked against their sources.
- px: a coding agent built around DeepSeek’s prefix cache
A 30-turn session costs $0.0149 off-peak. Most of the saving comes from never changing a byte the cache has already seen.
- What DeepSeek’s API won’t tell you
Notes from building px: cache hits at 2% of a miss, a model that spent its whole output budget thinking, images that vanish, and the $16 day that taught me to price a sweep before running it.
- Running a 27B coding model on my laptop
Qwen3.8-27B on an M4 Max: 20.8 tokens a second, a 20,000-token floor before any work starts, and the defaults that quietly broke it.
- Running Claude Code on a cheaper stack of models
A local LiteLLM proxy where a strong planner hands bounded coding jobs to cheap executors, with spend tracked per session.