These are the raw, unedited notes I wrote while shaping this article. They are closer to a brain dump than a prompt library: my thinking, philosophy, corrections, and direction. The ideas are mine. AI helped turn them into a smoother, more readable essay.

Research footprint

Open the evidence artifacts ↗
13human prompts
4agent sessions
9committed artifacts
112.5Mtokens processed

The included sessions span 38h 20m of elapsed wall-clock time, from to . The measured token total includes 112.1M input tokens (110.1M served from cache) and 398K output tokens. Reasoning tokens (95K) are already included in output. Elapsed time includes gaps between sessions and does not measure human hands-on work.

This is a provenance measure, not a quality score, cost estimate, or environmental-impact estimate. Token counts come from the private agent sessions identified in the public manifest, which records the log format, included session IDs, selected usage boundaries, and SHA-256 prefix commitments. The author can audit the totals against those logs; readers cannot independently reconstruct them without the private logs. The committed artifacts show the resulting work.

Prompt 1
what is the new agentic era of personal blogs and articles in september 2026? i have really little readership, and I am stuck... i want to have more
Prompt 2
are you suggesting a particular article?
Prompt 3
lets write this: Which AI fetchers send which headers, measured on a live site. And start the worklist for the following: 1. Pick one lane for the next ten posts. The obvious one is already in your catalog: agent-native publishing and honest first-party analytics for independent sites. Posts 016, 019, 020, and the ADR-0016 line are the most original thing you have. Nobody else is measuring agent readers with this rigor. Own it.
Prompt 4
are you following the "The doorway paragraph" is your own rule from the opening-discipline note: the first paragraph has to give a stranger the significance and the takeaway before any background. For this post that would be the one paragraph a cold reader can use without knowing who you are.?
Prompt 5
how to make this more valuable, validated, authoritative, legitimate, credible and so much more so that it gets referenced more?
Prompt 6
look up authoritative references on open source (do it yourself), then validate further and capture all this in the worklist
Prompt 7
The device-detector check used a Python parse of its YAML, not the PHP library. What? why not typescript
Prompt 8
then the four missing assistants probed with your accounts?? why can't we do it without accounts?
Prompt 9
cant you query the traces from cloudflare via cli? why are you using the chrome browser?
Prompt 10
i dont want the empty cells in the table
Prompt 11
btw, which ones do you want me to validate manually? and how? tell me if i can help i will help
Prompt 12
grok responded with http status 404, main heading this page doesnt exist, but i was on wifi
Prompt 13
afterall i want the researchFootprint to be measured and added to the article