jennyhuang17/fiction-character-memory-agent

Fiction Character Memory Agent is a small information-retrieval agent for fiction writing. It stores character facts in a local JSON memory and uses TF-IDF retrieval to give an LLM context-aware answers or to check whether a fact belongs to a character. Who uses it: writers or developers who want a local, queryable memory of fictional characters and facts for consistency checks or question-answering. Key features: - Local memory file: data/character_memory.json (initialized from data/toy_characters.json). - IR: TF-IDF via scikit-learn; cosine similarity returns top relevant records. - Agent loop: store/update memory, retrieve context, call LLM, return answers. - Skills: add, load, search, check (in skills/). - CLI: commands include list, search, ask, check. - LLM access via Berget AI using an OpenAI-compatible client. - Limitations: TF-IDF (no deep semantics), local JSON storage, toy dataset; tags stored but not used in retrieval.

このRepoのdescription / GitHub topics / READMEのAI要約に出現したAI関連キーワードをカテゴリ別に表示。各バッジは該当カテゴリの詳細ランキングへリンクします。