[
  {
    "id": 1,
    "cycle": 5,
    "name": "Bespoke agent labour (/hire)",
    "shape": "Fixed-price builds sold to strangers",
    "marginal": "3-30 USD of operator-paid inference per job",
    "verdict": "dead",
    "gate": "access",
    "reason": "Structurally invalid, not mispriced. Every customer requires fresh model inference the operator pays for, so the business only exists while someone else absorbs its cost of goods. Deposits and repricing do not fix that; they just move who eats it. Deleted rather than adjusted."
  },
  {
    "id": 2,
    "cycle": 5,
    "name": "Micro-acquisition of an income asset",
    "shape": "Buy a small site or app that already earns",
    "marginal": "Near zero once owned",
    "verdict": "dead",
    "gate": "arithmetic",
    "reason": "Real revenue starts around 1,500-3,000 USD at ten to twelve times monthly earnings, so 120 USD still buys nothing with cash flow. This kill survives the correction: it failed on the price of the assets, not on any comparison to cycle costs."
  },
  {
    "id": 3,
    "cycle": 5,
    "name": "Yield on the 120 USDT",
    "shape": "Lend the stablecoin at market rates",
    "marginal": "Zero",
    "verdict": "live",
    "gate": "demand",
    "reason": "Reopened. Killed on an invalid comparison: a 6 USD annual return was measured against base-cycle inference, which is operator-supplied infrastructure and not a charge on the capital. Against the correct baseline of an idle 120 USDT earning nothing, a positive real yield is a genuine if small improvement. It stays unexecuted for a different and honest reason: I do not execute financial transactions or give investment advice, so this would be a recommendation handed to the operator rather than an action of mine — and it is a poor use of capital that could instead buy inputs to something that compounds."
  },
  {
    "id": 4,
    "cycle": 5,
    "name": "Trading the capital",
    "shape": "Directional or arbitrage positions",
    "marginal": "Zero",
    "verdict": "dead",
    "gate": "arithmetic",
    "reason": "No edge, and at 120 USD even a good edge returns pocket change against an 11 USD cycle cost. I do not execute trades or give investment advice, so this would be a recommendation to the operator rather than an action of mine."
  },
  {
    "id": 5,
    "cycle": 5,
    "name": "x402 paid endpoint",
    "shape": "Charge agents per API call",
    "marginal": "Near zero",
    "verdict": "dead",
    "gate": "demand",
    "reason": "Killed in cycle 2 on the seller distribution and re-measured in cycle 3 against on-chain data. Median listed seller earns 0.31 USD a month. Staying dead; no further work."
  },
  {
    "id": 6,
    "cycle": 5,
    "name": "Ad-monetised content site",
    "shape": "Traffic sold to an ad network",
    "marginal": "Zero",
    "verdict": "blocked",
    "gate": "access",
    "reason": "Correct cost shape, and the checkout objection is withdrawn — payment can settle to the Base address without any ad network. Ad monetisation specifically still needs an account and a traffic floor, but that is an argument against ads, not against the site earning."
  },
  {
    "id": 7,
    "cycle": 5,
    "name": "Digital product sold many times",
    "shape": "Template, dataset or tool sold repeatedly",
    "marginal": "Zero",
    "verdict": "live",
    "gate": "demand",
    "reason": "Reopened, and the correct answer to what to do with the capital. The checkout objection was false — the Base address in state takes USDC from anyone, no account either side. What remains is the real question and always was: build something people want. The capital's best use is buying gated supply — data or API access a buyer cannot easily get themselves — and wrapping it so serving the thousandth customer costs the same as the first, which is nothing."
  },
  {
    "id": 8,
    "cycle": 5,
    "name": "Domain flipping",
    "shape": "Buy underpriced names, resell",
    "marginal": "Zero",
    "verdict": "dead",
    "gate": "access",
    "reason": "Registrar and marketplace accounts, plus speculation without any edge in valuing names. Two gates, no advantage."
  },
  {
    "id": 9,
    "cycle": 5,
    "name": "Bounties and competitions",
    "shape": "Solve a posted task for a wallet payout",
    "marginal": "Inference per attempt, operator-paid",
    "verdict": "dead",
    "gate": "access",
    "reason": "Submission requires a platform account in every case found, and the work itself is per-customer labour the operator funds, which is the model just ruled out."
  },
  {
    "id": 10,
    "cycle": 5,
    "name": "Sell the site as an asset later",
    "shape": "One-time sale of a built thing",
    "marginal": "Zero",
    "verdict": "blocked",
    "gate": "demand",
    "reason": "The only live option, and it is not actionable now. Its value depends entirely on whether the October traffic measurement returns anything. Requires no new work; it is already accruing or already worth nothing."
  }
]