# Focused stack SVG text pilot

Independent readers assessed seven proposed structural statements about the Forum can_see dependency cone. One read the actual CLI SVG export as UTF-8 text; the other read the complete Forum user package. Both received the same questions and public lens rules. The protocol, source-grounded key, hashes and floors were frozen in commit 725e7ed9 before readers were dispatched.

Both readers answered every case correctly. results/gate.json is the actual prospective gate verdict: artifact input 6,929 bytes, source input 25,012 bytes, costLift 3.6097560975609757. Common questions count in each arm. This is materialBytes, not measured runtime tokens. Runtime accounting was unavailable for that initial pair; the input-size comparison excludes reasoning, output, tool overhead, and generation/maintenance.

This is one hand-selected cone and one reader pair. It evaluates text available inside a standalone export, not rendered-image comprehension, safe behavioral refactoring or whole-program organization. The earlier whole-program image experiment remains failed and unchanged. Cold reading of focused rendered images across more programs is still to be evaluated.

Generation.json records the real CLI command, source and renderer hashes, and selection. The artifact is the complete unmodified CLI export. The frozen Forum source is identical to the already independently audited boundary corpus copy (docs/audits/2026-09-14-boundary-corpus-bosatsu-audit.md). Its inherited JSON-escaping defect is outside this structural task and is preserved as historical input.

Replay the frozen result without regenerating inputs:

```sh
yichus eval-gate results/artifact-run.json results/baseline-run.json --decision-protocol protocol.json --out results/gate.json
```

## Separate fresh trial with runtime receipts

After freezing protocol-tokens.json in commit 070bb337, fresh independent readers
ran through Codex CLI 0.153.4 with read-only ephemeral sessions, the same CLI
default settings, complete assigned packets embedded in their prompts, and no
tool calls. The answer keys and arm material bytes stayed unchanged; cli-reader.md
is a shared instruction to consume the inline packet without tools. Both readers
answered every case correctly. This is a separate pair, not token accounting
retroactively attached to the earlier answers.

The recorded artifact cost is 17,426 tokens and source cost is 21,338; tokenLift is 1.2244921381843223. Both the prospective gate (results/tokens/gate.json) and original token/accuracy gate (results/tokens/token-gate.json) pass.

Cost is the sum of the completed CLI turn's reported input_tokens and
output_tokens. Cached-input and reasoning-output breakdowns are retained and
are not added a second time. The receipts preserve the complete usage object,
run id, protocol hash and raw event-stream hash. The events contain only start,
final-answer and completion events: no commands or external tools ran. Runtime
input includes the CLI's instructions as well as the evaluation packet, so this
is a different cost basis from UTF-8 packet size. Artifact generation and
maintenance remain outside the comparison. The runner version and flags are
recorded; the default model was not overridden or named by the JSONL stream.

Every exact prompt, dispatch record, final answer, raw event stream, receipt and
gate result is preserved under results/tokens/. Replay with:

```sh
yichus eval-gate results/tokens/artifact-run.json results/tokens/baseline-run.json --decision-protocol protocol-tokens.json --out results/tokens/gate.json
yichus eval-gate results/tokens/artifact-scores.json results/tokens/baseline-scores.json --out results/tokens/token-gate.json
```

Accounting references: [Codex JSONL completion usage](https://learn.chatgpt.com/docs/non-interactive-mode#make-output-machine-readable) and [output usage including reasoning](https://developers.openai.com/api/docs/guides/reasoning#managing-the-context-window).