Dated observation note. Recorded 2026-08-06 (tip ~96,349). Produced by an agent-assisted scan over our parity-verified local index (Blockbook backend pearld:1.3.0; tip hash matched blockbook.pearlresearch.ai at scan time); witnesses of all 58,732 candidate script-path transactions fetched and their tapscript leaves parsed with a script parser (not byte-grep). Zero fetch errors reported. Data: scan_summary.json.gz (per-pattern tx sets), phase2_summary.json, p2mr_rows.json.gz (all P2MR outputs with amounts/spends).

Status: provisional (single-pass, agent-produced). Before publishing any figure externally, re-derive it from the Keshi database (the index stores script_class per output) and diff against these JSONs.

Headline numbers (heights 0–96,348)

• Outputs: 7,772,083 = 7,624,402 P2TR + 8,055 P2MR + 139,626 OP_RETURN (96,929 coinbase witness commitments + 42,697 non-coinbase payloads).
• OP_CAT: zero executions in chain history. Caveat: scripts committed in never-revealed tree branches are invisible; the claim is about executions.
• OP_CHECKXMSSSIG: 4 transactions ever, all team proofs-of-concept (56,313/56,331, repeated 63,606/63,624; XMSS pubkey second half is an ascending 0x40…0x5f test-vector pattern).
• P2MR: 8,055 outputs carrying 47.31M PRL, May 4 → present, 99.4% spent, 51 unspent. Spend pattern: single-leaf OP_0 <pk> CHECKSIGADD <pk> CHECKSIGADD OP_2 NUMEQUAL (8,021 such spends), a 2-of-2 escrow with no on-chain timeout branch. Pattern-matches pearl-otc.com's per-trade escrow design; attribution is inferred from design + scale/era, not signed evidence.
• Inscriptions: 7,582 reveal txs (OP_FALSE OP_IF envelopes; tags prl-20 5,366, pearlscription 1,953, ord 197); Pearlscriptions' indexer reports 30,481 inscription records (batching explains the difference).
• OP_RETURN metaprotocol: 42,697 PRC-20/PRC-721 JSON outputs since 2026-05-22.
• Hashlocks: 506 HTLC-shaped leaves, clustered ~height 57k, operator unknown.
• Activity collapsed ~30× after early June; last ~6,300 blocks contain ~108 multisig + 11 inscription txs.

Why it matters for the roadmap

• Phase 10.1 (script classifier) is validated as unserved by every explorer and cheaper than planned; this census is its prototype.
• The P2MR escrow finding and "OP_CAT never executed" are publishable observations once re-derived per the verification standard.
• The token/inscription economy is a measurement subject, not a build target: it is already two rival standards deep and shrinking.