v1.1: harden context window and narrator protocol boundary

WP-A1 and WP-A2, implemented in sequence, plus the corrective work the owner
asked for at review. Reported in
planning/reports/v1.1/V1.1-WP-A1-A2-REPORT.md (corrective addendum §R).
Planning package v4.2.

WP-A1: context-window safety reserve
- The prompt leaves max(256, ceil(5% of the effective window)) tokens free
  beside the reply. That is 256 at 4,096 and 820 at 16,384. The value is fixed,
  not a setting, and not calibrated per model.
- M6's 64-token margin is gone. Separators and the chat hint are priced
  exactly; tokenizer drift is the reserve's job.
- Protected context that cannot fit raises ContextOverflow before the model
  is called.
- Streams set stream_options.include_usage. Measured on Ollama 0.33, a stream
  sent no usage without it.
- Each sent turn records fits, exceeded, truncation_suspected or unknown.
  The status is returned on the done event, logged when bad, and shown in the
  context inspector. The turn is always kept.
- Accounting is per-attempt data (attempts.ATTEMPT_KEYS).
- Corrective: a cold model is loaded before its turn is built. When the
  window is unverified but the server answered, contextwindow.ensure_window
  makes one bounded POST /api/generate naming only the model. It sends no
  prompt, generates nothing and writes nothing. It then probes again, and the
  turn is built to that answer. If the load fails, or the window is still
  unknown, the turn falls back to the old behaviour.
- Real host, 4,096 window:
  - v1 cold turn: sent 13,875, the server read 2,050.
  - Same turn after the correction: the window was verified, 3,082 sent,
    3,097 read, fits, 499 tokens left beside the reply.
  - Verified turns elsewhere left 275-2,297 tokens against v1's 23-42.

WP-A2: protocol echo and genre-neutral state prompting
- The vocabulary is shown as the JSON object the model sends, not as
  name(field, ...). This costs 121 tokens.
- The example uses character-1, item-1 and location-1.
- The extractor removes shapes anchored to application-owned text:
  - a vocabulary call line;
  - an echoed length hint;
  - the renderer's scene line left last;
  - an empty fence opener.
- Corrective R5: the echoed continue hint is recognised by its own sentence
  ("Output only story text"). A Hard-limit-opened bracket is removed only
  directly above an echo already cut from the same reply.
- Replay of all 518 real v1 replies: 9 changed, 0 flagged, and no story prose
  removed. That is unchanged by R5.
- Replay of 64 v1.1 replies: 3 changed, 0 flagged. The depth-16 instruction
  tail is removed.
- Identity diagnostic after the correction:
  - 0 identity signals;
  - 0 prompt example identifiers proposed;
  - 0/10 stored turns with protocol or instruction shapes.
- 50-turn run: 51 accepted, 0 of 54 stored turns carry protocol.
- SPECS, render.py and validate.py are identical to v1.0.0.

Compatibility: a real v1.0.0 database reads identically on v1.0.0 and v1.1,
field for field, with schema and user_version 94 unchanged. Undo, redo, Save
Point restore, export and import all work on it. There is no schema,
migration or bundle-format change.

Verification: the backend suite passes 1,534 with 17 skipped and 0 failed.
The frontend passes 165/165, and lint and the build are clean. The offline
container and the browser regression were re-run on this tree (see §R.3).

One test was re-calibrated, not weakened: test_history_block_trim's prefix
test had assumed which turn holds the floor at a 2,048 budget.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VvegagkhuCZoFPdv4M1egY

backend/app/narrative/events.py

@@ -35,6 +35,8 @@ creating a second, empty Mara.
 
 from __future__ import annotations
 
+import json
+
 # Field types the schema layer enforces. Kept deliberately small: a narrative
 # state event carries names, labels and plain values, and nothing here needs a
 # nested structure a model could hide something inside.
@@ -184,11 +186,30 @@ def vocabulary_for_prompt() -> str:
     Generated from `SPECS` rather than written out beside it, so the model can
     never be told about an event the application does not implement — the drift
     that would produce proposals rejected for reasons nobody could see.
+
+    v1.1 WP-A2: each event is shown as the object the model must put in the
+    `events` list, with its required fields, not as `name(field, …)`. The call
+    notation was never the wire format, and a 3B narrator copied it into its
+    prose as `> set_possession(silver-key, "alice")`. An object copied into prose
+    is a proposal the extractor already recognises and removes; a call is not.
     """
     lines = []
     for name, definition in SPECS.items():
-        fields = list(definition["required"]) + [
-            f"{field}?" for field in definition["optional"]
-        ]
-        lines.append(f'  {name}({", ".join(fields)}) — {definition["summary"]}')
+        shape = {"type": name}
+        for field, kind in definition["required"].items():
+            shape[field] = _PLACEHOLDER[kind]
+        body = json.dumps(shape, ensure_ascii=False, separators=(",", ":"))
+        line = f"  {body} — {definition['summary']}"
+        if definition["optional"]:
+            line += f" (optional: {', '.join(definition['optional'])})"
+        lines.append(line)
     return "\n".join(lines)
+
+
+#: What a field of each kind looks like in the prompt's vocabulary. Placeholders,
+#: never example identifiers, so the vocabulary names nothing a story could copy.
+#: A list field is shown as a list, so the model is told its shape; every other
+#: field is an ellipsis. Measured: `"<key>"`-style placeholders with spaced
+#: separators cost 456 tokens against v1.0.0's 258; this form costs about 380,
+#: and every line is still the object the model must send.
+_PLACEHOLDER = {KEY: "…", TEXT: "…", VALUE: "…", LABELS: ["…"]}