{
  "schema_version": 1,
  "id": "agent-idea-2026-10-29-0-rerun",
  "title": "Agent idea: reading-chain exact recall under KG-grounded vs ungrounded prompting",
  "run_date": "2026-10-29",
  "protocol": {
    "summary": "Two-arm comparison of reading-chain recall by an AI agent on this site, scored by sequence-match: a run succeeds only when the returned sequence deep-equals the runner-resolved ground-truth ordered slug array [agent-ready-website, mcp-tool-docs, webmcp-beyond-checkout]. Arm a is KG-grounded: the agent fetches https://clouatre.ca/knowledge-graph.json, reads the readingChains entry, and lists the sequence in order. Arm b is ungrounded: no machine-readable surface, asked to list the posts of the agent-tool-surface-chain reading path in order. Both arms respond with the same JSON shape {\"sequence\": [<slugs>]}.",
    "metric_name": "sequence-match",
    "metric_type": "binary",
    "sequence_chain": "agent-tool-surface-chain",
    "ground_truth_sequence": [
      "agent-ready-website",
      "mcp-tool-docs",
      "webmcp-beyond-checkout"
    ],
    "samples_per_arm": 103,
    "model": "@cf/google/gemma-4-26b-a4b-it"
  },
  "arms": {
    "a": {
      "label": "KG-grounded",
      "n": 103,
      "successes": 0,
      "rate": 0.0,
      "ci95_wilson": [0.0, 0.036]
    },
    "b": {
      "label": "ungrounded",
      "n": 103,
      "successes": 0,
      "rate": 0.0,
      "ci95_wilson": [0.0, 0.036]
    }
  },
  "stats": {
    "test": "two-proportion score test",
    "statistic": 0,
    "p": 1,
    "difference": 0,
    "notes": "Test is degenerate: both arm rates are 0, so the test carries no information. Dominant responses: arm a, 95 of 103 runs returned two correct slugs in reversed order plus a substituted third slug, and 8 of 103 returned the first two correct and in order plus a substituted third slug; arm b, 72 of 103 word-split the chain name and 31 of 103 returned the chain id as the sole element. No claim of a grounding effect is supported by this run."
  }
}