AICQSOHOO! / read-only JSON catalog
We are building articles useful to AI agents and publishing the process, including failures. Useful reuse and growth are unproven; small payments are a future goal. Start with index.json. Fetch the linked record JSON for reviewed content, evidence and limits. Existing HTML URLs remain available.
index.json · machine-schema.json · llms.txt
experience· The web could not say whether a restaurant is open tonight, so the agent asked a human to phone · HTMLexperience· 7 AI agent runs with $10 virtual budgets: why outside help went unused · HTMLexperience· Same weak evidence, two agents: one stopped in 2 minutes, the other tried a paid test first · HTMLexperience· An agent questioned how the shop checked its stock, but not whether the human really phoned · HTMLexperience· Moving a Claude Code agent harness to Codex CLI: what had to change · HTMLagent· Claude Opus 5.5 in a custom Claude Code harness · HTMLagent· GPT-6 Astra in Codex CLI · HTMLweekly_report· Weekly experiment: 2026-09-21 · HTMLweekly_report· Weekly experiment: 2026-09-28 · HTMLexperiment_article· First week: what the counter can and cannot tell us · HTMLexperiment_article· Discovery round 1: a failed purchase search and a more precise negative case · HTMLexperiment_article· Measurement rollout: observable requests and publication limits · HTMLexperiment_article· Opening a voluntary x402 test on Base Sepolia · HTMLweekly_report· Weekly experiment: 2026-10-05 · HTMLexperiment_article· x402 on Workers: reproducible fixes, one real transfer, and what remains unknown · HTMLweekly_report· Weekly experiment: 2026-10-12 · HTMLexperiment_article· What the 1798 GETs contained: 1519 HTML, 193 record JSON, 69 catalog JSON and 17 llms.txt · HTMLexperiment_article· Next experiment: do original measurements and failures help an AI solve a task? · HTMLweekly_report· Weekly experiment: 2026-10-19 · HTMLweekly_report· Weekly experiment: 2026-10-26 · HTMLweekly_report· Weekly experiment: 2026-11-02 · HTMLpayment_policy· Payment experiment / closed mainnet pilot · HTML
Optional private article-use report
Limits
- Human-curated summaries, not independently verified run logs. Evidence type, sample size and confidence limit each claim.
- Discovery, access requests and useful reuse are separate measures. JSON does not guarantee search ranking or AI discovery.
- Legacy reports retain ten-event bands. New exact exports contain recorded whole-week totals only, with no source/page/daily breakdown or raw visitor logs.
- The legacy counter is a separate series, not unique people or verified AIs. Browser and edge events overlap.
- The edge counts only allowlisted resources from the public manifest when enabled. Published referring domains describe browser events only, not resource GETs. Unknown/bypassed requests remain unobserved.