The canon-log schema and NewGame code already migrated to race_id/
calling_id + allowed_callings, but two copies of the roster outside
that schema were missed, and no parity test guards them:
- docs/canon-log.md (the cross-boundary contract doc) still documented
the OLD player shape and allowed_classes. Anyone building a player
block from the doc would get a 422.
- .claude/skills/world-building's schema reference still emitted
allowed_classes with two dead calling ids (assassin, priest). Author
a new origin with that skill and it fails origin.schema.json
(additionalProperties: false, allowed_callings required) AND, if that
ever loosened, silently allows zero callings at runtime.
Also:
- add schema tests rejecting a dead class_id field and an unknown
calling_id (priest)
- guard NewGame._validate's container types (spend/skills) so a
malformed JSON round-trip (null spend, string skills) produces a
front-loaded error list instead of a GDScript runtime crash
- extend the no-mechanics-in-content test to db.races, not just
db.callings
- rewrite the roadmap's M4 bullet: the two contract migrations landed;
only the creation screen and title->creation->shell flow remain
client: 250/250. api: 74 passed, 2 skipped. content_build --check: clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QYa9u7Kdxv5gX4AnwWexy8
Three callers (main_window_shell.gd, narrate_harness.gd, npc_harness.gd)
still built LogPlayer with the OLD 3-arg shape (name, class_id,
luck_descriptor) after LogPlayer.new() was migrated to the 4-arg
(name, race_id, calling_id, luck_descriptor) shape. This was invisible
to both grep and the test suite: every ctor param carries a default, so
GDScript compiles a 3-arg call against a 4-arg signature without error —
it just silently mis-binds positionally. "sellsword" landed in the
race_id slot (rejected — not a race), the luck descriptor landed in the
calling_id slot (rejected — not a calling), and luck_descriptor itself
fell back to its default "". Every field in the emitted dict came out
empty, which 422s against the canon-log schema's minLength/enum
constraints. Grepping for class_id can't see this, because there is no
class_id token left anywhere — the bug is purely positional.
Also fixes prompts.py's _article(""), which returned "an" because
Python's "" in "aeiou" is True, producing a doubled-space/wrong-article
digest line when race_id is empty. _describe_player now builds its
descriptor from whichever of race/calling are present and omits the
clause entirely when both are absent.
Adds a schema-parity guard for origin.schema.json's inlined
allowed_callings enum, which duplicated the calling roster with no
runtime check tying it to Callings.IDS. Renames leftover "class"
vocabulary in test names/comments to "calling".
Regression coverage:
- client/tests/unit/test_entities.gd: ctor populates both race_id and
calling_id; a calling passed into the race_id slot (the exact shape
of the three broken call sites) yields an all-empty row.
- api/tests/test_prompts.py: empty-race and empty-race-and-calling
digest rendering, asserting no doubled space and no dangling article.
- client/tests/unit/test_schema_parity.gd: origin.schema.json's
allowed_callings enum matches Callings.IDS.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QYa9u7Kdxv5gX4AnwWexy8
class_id is the rulebook's word; the world has callings — that was the whole §17
reconcile. The contract the Narrator reads should say what the world says. There
are no saves (M9) and no deployed clients, so this is as cheap as it will ever be
and strictly more expensive every milestone after.
race_id reaches the log because the AI must be able to describe the player:
api/app/prompts.py rendered 'a sellsword' and now renders 'a beastfolk cutpurse'.
It humanizes the id (the model must never read 'hedge_mage') and picks the right
article ('an elf', not 'a elf').
LogPlayer's hardcoded CLASSES const is deleted — the roster is Callings.IDS. A
parity test reads the JSON Schema from the client and asserts its enum equals the
code, so the two sides of the HTTP boundary cannot drift.
The deserter allowed three callings that no longer exist. He now allows all seven;
a thematic gate is a content decision for the Greywater authoring.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QYa9u7Kdxv5gX4AnwWexy8
- call_log._default_write: swallow any exception from the default file/
stdout sink and report to stderr, so a log-write failure (disk full,
bad permissions, misconfigured CALL_LOG_PATH) never turns a successful
narration into a 500 (charter §13). Injected write= sinks (used by
tests) are left to surface their own errors.
- ollama_client.chat: catch ValueError alongside httpx.HTTPError in the
retry loop so a 200 response with a non-JSON body (JSONDecodeError is
a ValueError subclass) counts as a failed attempt and falls through to
ModelError after the one retry, instead of escaping chat() uncaught
(charter §12 — one retry, then ModelError, nothing else escapes).
- main.py: update stale docstrings — /dm/narrate is now fully wired
(routing + model call + logging); the other four roles remain stubs.
Regression tests added for both fixes (TDD: watched RED, then GREEN).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
/dm/narrate now makes a real Ollama call (Task 7). The role-endpoint
loop tests in test_endpoints.py had no mock, so every default pytest
run was silently hitting a live Ollama instance on localhost:11434 —
breaking the hermetic-suite constraint. Add an autouse fixture that
stubs narrate.ollama_client.chat and narrate.call_log.record so all
five endpoints stay hermetic; assertions (200 on valid, 422 on
invalid) are unchanged.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- canon_log.py: stringify jsonschema error-path segments before sorting so a
future schema mixing array indices and object keys can't TypeError (#1)
- main.py: RequestValidationError handler reshapes pydantic body failures into
the same {"detail": {"canon_log_errors": [...]}} envelope as schema failures,
so the client parses ONE 422 shape (#2a) + tests
- pytest.ini: filter the starlette/httpx TestClient deprecation → pristine output (#4)
- docs/canon-log.md: living contract for the client (Plan B) — schemas, the §7
boundary, the unified error envelope, per-role injection (#3)
Full suite 26 passing, 0 warnings.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
