Poking Claude With Sticks Until It Gets On With It — Overview
Audio deep dive: teaching Claude Code to stop asking permission — blocker laundering, menu failure, fire doors, and a ledger that reopens on retraction.
Audio deep dive: teaching Claude Code to stop asking permission — blocker laundering, menu failure, fire doors, and a ledger that reopens on retraction.
Audio deep dive: operating Home Assistant entirely over its APIs — three auth channels, reload semantics, and why exit 0 is not evidence.
A NotebookLM deep dive across all twelve essays: legibility as governability, and opacity as the freedom worth keeping — the four movements in one listen.
Audio deep dive into the world of The Ungovernable Body — Continuity, the Twin, the Halo, and the right to remain uncounted, from a 25,000-word codex.
Audio deep dive into an agent-operable film studio — canon as YAML, decision records, machine-validated dialogue, and a feminist techno-thriller short.
A present haunted by its own archive. On feeds that never end, time folded back on itself, and the loss of the past as past.
Deletion as ritual rather than accident. On ceremonial erasure, controlled burns, and what it would mean to grieve data properly.
Audio overview of ClawdCraft — Claude Code as an in-game creature kids talk to, and why its hard limits live in code instead of the prompt.
Audio overview of Dead Air — a failure-eval harness for voice agents, and why the worst failures rarely live in the transcript.
Audio overview of Factory Floor — turning a from-memory 3D-printing habit into a validated, human-gated pipeline with a single source of truth.
Audio overview of Understory — a rigor lab stress-testing stacked income streams on Tasmanian conservation land, where every claim is tagged and gated.
The database self remembers what you do not. On contradictory records, leakage, and who controls the story your data tells about you.