# CBE-IX public method profile

## Source snapshot

- Source: author-provided `Capability Boundary Expander - Iterative Exhaustion AI Execution Prompt`.
- Source created/updated: 2026-08-24.
- Supplied source SHA-256: `7df12892df3b1f7dc8e1fced3c833110ce93ec3ade77819c3bdc659b198b9953`.
- Projection prepared: 2026-09-27.
- Source status: supplied project document. No public canonical repository locator was established for this prompt. The source bytes are not republished here.

## Independent source verification

The exact author-provided source was available to the implementing reviewer and its SHA-256 was recomputed on 2026-09-27; it matches the value above. A reviewer with only this public repository cannot independently recompute that source hash. Obtain the named source document from the author and hash the original bytes before using the section mapping below. The profile and this receipt remain derived projections, not replacements for the prompt.

## Claim mapping

| Public claim | Source section |
|---|---|
| Semantic fixed point | Fixed-point model |
| Target-only queue, separate unresolved decisions | Expansion frontier; CBE-I3, CBE-I5, CBE-I7; Step 7 |
| Persistent discovery and classification history | Iteration ledger; CBE-I7 |
| Seven exhaustion checks | Step 11 - Convergence test |
| External evidence/authority question routed to residual or blocker | C5 in Step 11 |
| Repeated equivalent candidates without progress | Step 12 - No-progress / cycle detection |
| Aggregate identity check after every material delta | Step 13 - Identity drift test |
| Eight exact stopping names and peer-takeover risk | Terminal states |
| Preserve ledger for new evidence | Stop / re-entry conditions |
| Complete ingestion responsibility baseline | Base CBE prompt, Example 2; see the [CBE source snapshot](../cbe/README.md) |

## Teaching additions

The ingestion example is adapted from CBE Example 2. The reader-assistant rejection, unknown regional-access ownership, explicit missing/conflicting/unrecognized-classification handling, ambiguous parser matching, accumulated reader-platform drift and all narrated iteration traces are curated synthetic examples. They are not historical deployment evidence. The narrow format-detector fixture is a separately declared target. I1 explicitly replays the base CBE snapshot, including duplicate detection and safety-gate coordination with a delegated Malware Scanner. The identifier comparison, provided ingestion record set, new/duplicate submission rule and clear/flagged/unavailable scanner contract are finite teaching assumptions that make the narrated I1 closure inspectable. The scanner owns detection; ingestion preserves its answer and holds submission unless clear. I2 closes new gaps about ambiguous parser matches and conflicting classification returns. The alternative drift case retains its unprocessed initial responsibility areas rather than claiming the successful trace's closure.

The CBE lab compares a reader prediction with a curated answer. The CBE-IX lab walks fixed traces; it is not a semantic convergence engine. Different terminal cases are alternatives, not a fixed precedence algorithm. No API, model, peer operation or repository write is performed. No visitor input is uploaded or persisted.

## Verification

Run `python tests/site_copy_and_canvas_lint.py`, serve the repository locally, then run `node tests/cbe_labs.mjs`, `node tests/solution_explorer.mjs`, `node tests/method_interactions.mjs` and `node tests/blueprint_authority_labs.mjs`. Review desktop/mobile screenshots and the no-JavaScript fallback. Existing method examples and lab logic remain available.