Sessions and techniques

How I work

See complete work sessions, including mistakes, corrections, and the evidence behind each decision.

Open a full session or jump to one tested technique you can reuse.

12Sessions 9Techniques

Session 02

Start with a complete sessionThe Browser Is a Shell CommandOpen the session

01See the work as it happened.Follow the decisions, failures, and corrections in order.

02Reuse a tested technique.Take one useful part without replaying the whole session.

Complete collection

Browse sessions and techniques.

Search all 21 records, or open either collection directly.

21 of 21 sessions and techniques in view

All sessions and techniques

Full sessions

12 sessions

See the work in the order it happened.

Published Instagram story rendered by the demo's pipeline

Full sessionTwelve Messages

Everything typed to take a live event from spreadsheet chaos to published social content — and the method that made the other 99% happen.

Open session

Demo 02 title slide: the browser is a shell command, with a real browse-shot transcript

Full sessionThe Browser Is a Shell Command

Replacing 18,000 tokens of always-loaded browser-automation schema with ten shell commands and a README — and why the agent gets sharper, not weaker.

Open session

Demo 03 title slide, featuring a real read-guard denial of an oversized screenshot

Full sessionTaught Once, Enforced Forever

Corrections given in chat decay when the session ends. The fix is changing the agent's environment — helpers, deny-hooks, and CI ratchets that carry the rule forever.

Open session

Demo 04 title slide: a real artifact-miner run over 2,747 transcripts

Full sessionYour Sessions Are a Corpus

A 24 GB session archive forced the next question: how do you keep the lessons, prove closeout, and let the raw record expire?

Open session

Demo 05 title slide with a real agent-lane ticket and its labels

Full sessionThe Product That Files Its Own Tickets

End users file feedback, an LLM judge triages it into GitHub Issues, an agent implements the safe ones — with the autonomy boundary in deterministic code and a human holding the only merge key.

Open session

Demo 06 title slide with the shipped-twice landmine exhibit

Full sessionThe Registry of Landmines

Some load-bearing facts are invisible to search — the code compiles and the behavior is still wrong. One registry file holds them, a derive catalog enforces them in CI, and a meta-test keeps the registry from lying about itself.

Open session

Demo 07 title slide with a claim-versus-runtime-check exhibit

Full sessionThe Agent Said It Checked

"Verified" is a sentence, not a fact. A security migration passed its audit while production was broken — and the discipline that came out of it caught three more false claims building this very series.

Open session

Demo 08 title slide: the staged pipeline with gate posts between every stage

Full sessionGates Between Agentic Stages

Agents are strong inside a stage and unreliable at the boundaries. A delivery methodology built on that: deterministic gates between agentic stages — 88 versioned revisions, 14 running initiatives, and the day production published fiction.

Open session

Demo 09 title slide: a README claim beside the code fact that contradicts it

Full sessionThe Beautifier Was an Auditor

A skill built to make a README prettier turned out to be a drift detector. Across five repos it flagged a deploy doc still pointing at a platform the project left months ago, a command list that no longer matched the actual commands, and capabilities the code had grown that nobody had written down. The visual polish was the smallest thing it did.

Open session

Demo 10 title slide with a confidence tag being downgraded from verified to search-derived

Full sessionThe Chiropractor's Four Questions

A sports chiropractor asked four plain questions about selling rehab programs from his website. The AI's answer was confident, tabulated, sourced — and its evidence was marketing all the way down. Twice.

Open session

Demo 11 title slide: the sycophancy was in the config, with the operator's verbatim question about model versus configuration

Full sessionThe Sycophancy Was in the Config

A rival model reviewing my agent's work looked sharper — it demanded competing candidates and restraint. So my agent A/B tested its own harness: same model, same prompt, config on and off. The vanilla arm pushed back on me twice. The configured arm never did.

Open session

Demo 12 title slide: one component I didn't already have, with the operator's verbatim request to compare the tool against what he already runs

Full sessionOne Component I Didn't Already Have

A plugin advertising eleven agents and fifty-four hooks, measured against the setup I already run. Of the eight components that could land on my machine, three were already going, one was the wrong platform, one was broken upstream, two were costs. The delta was one.

Open session

Techniques

9 techniques

Tested parts of the work you can reuse.

Applied techniqueRun the Subtraction Before You Install

A feature list is half a subtraction. Enumerate your baseline, sort the tool's list into three buckets, then search its tracker for the uninstall.

Open technique Applied techniqueBare-Arm Test Your Agent Config

A control arm for your own harness: bare mode, isolation arms, and rules validated before adoption.

Open technique Applied techniqueMine Your Own Transcripts

Treat your agent sessions as a searchable corpus — corrections become priors the next session reads.

Open technique Applied techniqueGate Every Hand-Off

Agents work inside a stage; deterministic gates verify between. Gate the seam, not the middle.

Open technique Applied techniquePromote a Correction to a Guardrail

Wrapper → deny-hook → CI ratchet: promote a repeated correction to the lowest rung that holds.

Open technique Applied techniquePut Provenance Labels on AI Research

Three label tiers, an adversarial re-derivation loop, and the honest downgrade — shown working.

Open technique Applied techniqueA Registry for What Grep Can't Find

Facts grep can't surface, held in one registry and enforced in CI — the conclusion, pre-paid once.

Open technique Applied techniqueClose the Session, Keep the Lesson

Turn a finished agent session into durable project truth and compact recall before its raw transcript expires.

Open technique Applied techniqueTwo Ways to Draw a System

Hand-authored SVG vs themed Mermaid — chosen per diagram, both shown working.

Open technique

Loading the complete collection…