5 Commits

Author	SHA1	Message	Date
veguAI	b7d65f6d9d	Card import: keep the card's attributes when attribute extraction produces nothing (#159) ... * Card import: keep the card's attributes when attribute extraction produces nothing (#158) * Simplify the empty-sheet guard and sync the scripting-wrapper contract (PR 159 review)	2026-07-26 20:09:55 +03:00
veguAI	da464bb5b0	Card import: keep the card's description and example dialogue when generation fails (#156) ... * Card import: keep the card's description and example dialogue when generation fails (#154) * Document the empty-description contract and scope the import fallback note (PR 156 review) * Scope the import fallback admonition title and link Fast Character Generation (PR 156 review round 2)	2026-07-26 11:24:30 +03:00
veguAI	25d038e8c9	Character creation refactor: unified flow + Fast mode (#138) (#139) ... * feat(creator): consolidated character generation (Fast mode) with per-aspect orchestrator * test(creator): consolidated generation unit + baseline template tests; strict anchor extraction for one-shot misses * feat(director): persist_character unified flow - Fast mode branch, generate=False manual mode, pre-generated field params * feat(load): route card import AI extraction through creator consolidated generation in Fast mode * feat(nodes): add agents/creator/GenerateCharacter node; PersistCharacter accepts pre-generated fields * refactor: remove deprecated WorldStateManager.create_character path; world editor manual creation uses persist_character * docs: changelog entries + creator settings docs for Fast character creation * fix(creator): parse consolidated response by section boundaries - real models write open sections without closing tags * fix(creator): drop bare name/header lines from consolidated example dialogue * fix: address PR #139 review - node chain description passthrough, manual-mode template gate, shared attribute parser, fast import error tolerance * fix: address PR #139 review round 2 - late-name renormalization guard, speaker-drop heuristics, cancellation propagation, universal name guard * feat(creator): one-shot token budget setting (1024-8192, default 4096) for Fast character creation * feat: consolidate attribute templates into the one-shot + review round 3 fixes (strict extraction, mixin action, name guards) * fix: address PR #139 review round 4 - conditional template fold, augment folding, response-budget-aware context reservation, budget clamping * fix: address PR #139 review round 5 - augment in fill-miss fallback, fold eligibility vs Consolidate set, clamped history budget, fold precondition docs * fix: clamp one-shot response budget to client context so history + overhead + response always fit * fix: address PR #139 review round 7 - docs-index entries for Character Creation pages, response-budget docstring bound * fix: template fold works with name determination (placeholder formatting); consolidate_templates becomes an 'Attribute templates' entry in the Consolidate flags * test: re-level character-creation test mocking to the client-response layer (#140) (#144) * feat(creator): richer one-shot instructions ported from standalone prompts; quick toggle for Fast mode * feat(creator): label fast mode 'Fast Character Generation' for the quick toggle * fix: propagate GenerationCancelled in split-mode card import helpers (#142) (#145) * refactor: decompose the character-generation flow (#141) (#146) * refactor(creator,director): decompose the character-generation flow (#141) * refactor(creator,director): pydantic request models; generate_character_unified takes the request object (#141) * refactor(creator,director): address PR #146 review - sheet visibility, node None-name guard, request hardening (#141) * refactor(creator,director): PR #146 round 2 - completion status parity, enforcement ordering, test pins (#141) * refactor(director): PR #146 round 3 - completion status placement, cap pins (#141) * fix: address PR #139 review - fold placeholder gated on determine_name, cancel propagation in card import pipeline, dedupe fast/split orchestration, orchestrator-side character detach, docs for Attribute templates fold * fix: address PR #139 review round 2 - sheet prompt gets pre-creation description via explicit context character, detach field-fidelity coverage, aspect type annotations, Fast Character Generation naming, extraction polish * fix: address PR #139 review round 3 - uniform context-character rendering in sheet prompt (skip_characters), augment detour comment restated, Consolidate description rename, test state/fidelity gaps * refactor: move StrictAnchorExtractor to talemate.prompts.response with the other extractors (PR #139 review) * feat(creator): default Consolidate to every aspect incl. example dialogue * feat(creator): render writing style instructions in the consolidated character generation prompt; reword Fast Character Generation description (no upper request limit, less detail as failure mode) * fix: address PR #139 review round 6 - example dialogue guidance reaches the consolidated one-shot, extractor docstring + unit tests, dialogue_examples event caveats, writing-style docs, changelog accuracy	2026-07-25 13:17:11 +03:00
veguAI	f38d330ba1	Fix node code bugs surfaced by the docstring audit (#115) (#122) ... * fix: node code bugs from docstring audit (#115) - runtime crashes, phantom/duplicate sockets, registry typo aliases, property default mismatches * fix: address PR #122 review - register template scene types on intent_state, StageExit node_state_pop, annotations, shared graph-test helpers, migration round-trip test * docs: correct the circular-import chain named in apply_to_scene's deferral comment * fix: SetScenePhase defaults to the shipped roleplay scene type instead of an empty string that errors on run	2026-07-20 17:28:56 +03:00
veguAI	89e01ac138	Node editor node reference glossary (#107) (#111) ... * docs: complete node editor node reference glossary (#107) - generated by new nodes.tools glossary subcommand, 47 categorized pages covering all 403 nodes, docstring gap fixes, help-agent docs-index entries * docs: fix Jinja2 Format discoverability - document dynamic-input template variables, move node to string page, link key inference from dynamic-input notes, strip links from index summaries * help agent: keep node reference out of the chat prompt - collapse glossary entries to one pointer line (~19k chars saved), agent retrieves node info via search_docs/read_doc tool calls * help agent: replace in-prompt docs index with find_docs lookup tool - prompt carries a 13-line section overview (~1.5k chars vs ~66k), pages located via keyword-scored find_docs at runtime * changelog: help agent prompt slimming + find_docs lookup * docs: sort node reference index and nav alphabetically by page title * review: move module constants into top-of-file constant blocks (glossary MAX_DEFAULT_LENGTH; docs SECTION_DESCRIPTIONS, _FIND_STOPWORDS, FIND_DOCS_LIMIT) * review: word-boundary find_docs scoring with plural/-ing folds, find_docs+section_overview unit tests, jinja2 template field description override * docs: help agent overview no longer claims an in-prompt index of all documentation pages * docs: full node docstring audit pass - 335 classes audited by 5 agents, ~89 accuracy fixes (nonexistent/misnamed sockets, wrong behavior claims), ~94 expansions, 5 factually-wrong PropertyField descriptions corrected, glossary regenerated	2026-07-20 01:49:33 +03:00