docs(examples): finish the DX-audit backlog — 5 examples polished, proven, simplified

[drewstone]

Clears the remaining backlog from the examples DX audit (#428). Each fix verified — tsc 0, lint clean, and the offline-runnable ones run end-to-end ($0).

Example	Fix	Proof
intelligence-drop-in	The headline (OFF tier = $0 intelligence) was asserted, never shown. Now records inference-only at OFF, flushes, and reads the exported span back to prove intelligence_usd=0 (throws otherwise). Dropped the confusing double-client ceremony; README rewritten (glosses Mode 0, adds "Going live").	runs $0, exit 0
strategy-suite	The authored doubleCheck was a near-clone of built-in refine. Rewrote it as a policy the built-ins don't have — require two consecutive passes before done (a flake guard) — the real point of defineStrategy. Fixed the README (listed 4 built-ins, ran 2).	runs $0, exit 0
agents-of-all-shapes	shipToTangleOtlp hand-rolled the OTLP envelope — the exact wire createOtelExporter produces (the primitive the example teaches). Routed through createOtelExporter (45→25 lines). npx→pnpm.	runs $0, exit 0
self-improving-loop	The n=3 "small-n mirage" warning was restated 4×; trimmed to one authoritative block + brief pointers.	—
agentic-data-creation	Softened an unverifiable future-dated citation (arXiv 2606.25996 / "Meta FAIR" / "the paper's Table 1" + specific figures) stated as fact → the real technique (self-instruct, Wang et al. 2022) + "illustrative target" framing.	—

The key catch (senior detail)

intelligence-drop-in's original assertion would have failed if I'd naively "read the span back" — recordOutcome doesn't clamp, so recording intelligenceUsd: 0.05 at OFF was itself wrong (OFF can't have intelligence cost). And the OTLP attribute is loop.tangle.usage.intelligence_usd with intValue: "0" (string), not tangle.usage.intelligence_usd. Both found by actually running it and dumping the wire payload — not by assuming.

Verification: src + examples typecheck 0; lint clean (359 files); intelligence-drop-in / strategy-suite / agents-of-all-shapes all run offline to exit 0.

[tangletools]

✅ Auto-approved drewstone PR — 8703e398

This PR was opened by the trusted drewstone account.
The full PR reviewer audit still runs separately and will publish findings if it detects issues.

_{tangletools · auto-approval · reason: drewstone_author · 2026-07-01T05:24:54Z}