1.2.0 (2026-04-19)
- bot: add finish reason tracking and separate error/failed counters (5b0d95c)
- views: add current/previous/best run display and status indicators (1d54796)
- config: add community config example (a245a0c)
- config: note that model seed is not universally supported (6269c59)
- update commit command with workflow and scope definitions (20282ff)
1.1.1 (2026-02-12)
- bot: add cache_control to strategy message (068ff87)
- improve game resilience and error diagnostics (2c7b48f)
- llm: catch JSONDecodeError in LLM retry loop (0a17bfa)
- update balatrobot to 1.4.1 (a05daa6)
- cli: add note about tasks vs workers and seed regex (ebfe844)
- envrc: add .envrc.example (ca44aef)
- improve docs and fix inconsistencies (60363e3), closes #62
- index: update logos to have same size and add underline (a127565)
- installation: add tool use warning (4b8acf9)
- mkdocs: change accent color to shades of orange (06d85b7)
- readme: add screenshots (2d7280c)
1.1.0 (2026-02-02)
- balatrollm: add views server (e1ccc05)
- cli: add BALATROLLM_CONFIG env var support (2fc5f36), closes #57
- collector: add support model name without
/(cd30b65) - strategies: add conservative strategy (0ae6aae)
- strategy: add aggressive multiplier-focused strategy (fc958ce)
- views: add task and responses views - WIP (446c388)
- balatrollm: remove hard-coded usage in the DEFAULT_MODEL_CONFIG (9f23461)
- config: add usage include to config/example.yaml (bce4c92)
- views: reduce memory usage by only parsing the last 10 responses (3c8481e)
- CLAUDE: add views flag in CLAUDE.md (428b07f)
- cli: add --views flag (686d9b3)
- index: add BalatroBench link (3aff5db)
1.0.9 (2026-01-18)
- strategies: set strict to false for all tools schemas (a313210)
1.0.8 (2026-01-14)
- executor: revert session_id change (7419cfc)
1.0.7 (2026-01-14)
- executor: remove reset client in finally block (c525272)
1.0.6 (2026-01-13)
- update balatrobot to 1.3.4 (5f1f031)
1.0.5 (2026-01-13)
- executor: reset game to menu state before returning port to pool (05a468c)
1.0.4 (2026-01-12)
- update balatrobot to 1.3.3 (af9ae33)
1.0.3 (2026-01-11)
- pyproject.toml: update balatrobot to 1.3.1 (a02e363)
1.0.2 (2026-01-11)
- update balatrobot to 1.3.0 (b354476)
1.0.1 (2026-01-09)
- strategies: unify pack tools into a single function (6fb0ba0)
1.0.0 (2026-01-06)
- add init.py to src/balatrollm (c8d350a)
- add executor to start/stop balatro instances and tasks (3f80e3f)
- add llm client (2c35963)
- add new config/task structure (10ab612)
- add strategy manifest class to strategy.py (fb91793)
- cli: add new CLI entry point for `balatrollm`` (f87ea11)
- client: add BalatroClient and BalatroError classes (a3500cb)
- collector: flatten the structure of the stats (2540725)
- config: change default model to gpt-oss-120b in example.yaml (8d94917)
- config: remove legacy config and add example config (76f0e37)
- executor: add error message if the task fails (f9ad726)
- executor: add log path to executor output (d10e026)
- executor: spawn instances in parallel (49b03b9)
- scripts: add balatro launchers for linux, macos, and windows (4b770eb)
- strategies: add
packendpoint support (3b3c487) - strategies: add schema to manifest.json (36ca7bb)
- strategy: add default strategy (0a8fb96)
- strategy: add strategy manager class (16bcc64)
- scripts: wait for port to be ready in launch script (3e4fc73)
- strategy: remove booster packs references (069a2a1)
- test target in the Makefile (a95e78e)
- add comments to collector.py about various ids (16b3cdb)
- assets: add balatrollm-white.svg (a1a80c5)
- move CONTRIBUTING.md to docs/contributing.md (bc037b9)
- remove order list and use bullets instead (928712b)
- update CLAUDE.md to reflect current state of the project (9bb83f8)
- update docs index with repos assets (78e0127)
- update mkdocs.yml following mkdocs from balatrobot (89eb755)
- update pages in the docs (26c8d3f)
- update repo link in CLAUDE.md (7301d78)
- update the README following the balatrobot README (9caa373)
- release balatrollm v1 (ec2fcb3)
0.16.0 (2025-11-03)
- finer metrics generation by balatrobench (0d684aa)
0.15.1 (2025-10-30)
- update balatrobot to 0.7.5 (3ae6b1a)
0.15.0 (2025-10-30)
- add webp conversion to balatrobench cli (8afe67c)
- add better documentation for strategies (1af7822)
0.14.1 (2025-10-28)
- add keepalive get_game_state to keep TCP connection alive (992faf3)
0.14.0 (2025-10-25)
- add logic to load/save strategy manifest (e55b885)
- add manifest.json files for strategies (14978aa)
- add strategy metadata to benchmark (88ebc44)
- configure balatrollm via env vars (03c0fd0)
- new cli for balatrobench (973bc3e)
- new separate command for benchmarking (b9a1284)
- update the aggressive strategy (e51b9ca)
- add video in the readme (328d9bc)
- fix list rendering using sane_lists (a52decb)
- remove old screenshots assets (ef831a0)
- rewrite docs from scratch (8c78a9b)
- update .envrc.example with all the env vars (8852bc5)
- update desc for llms-txt and set index.md as home (dbe4a95)
- update README.md (8ab0aa8)
- update titles in the README (27006f5)
0.13.2 (2025-10-24)
- add pypi badge (94b4fbf)
0.13.1 (2025-10-23)
- extract reasoning from tool calls if not available in response (0422061)
- update benchmark.py to handle multimodal content (97b5b04)
- improve conversion to AVIF (2e3be75)
0.13.0 (2025-10-23)
0.12.0 (2025-10-22)
- add --no-screenshot flag to balatrollm (d15f087)
- add --use-default-paths flag to use balatrobot's default storage paths (0d47320)
- add rearrnge jokers tool to default strategy (64016d9)
- use prompt caching (2315919)
- use throughput instead of price for sorting (63d4b01)
- send message get_game_state to trigger frame rendering (8e89521)
- explain --no-screenshot --use-default-paths flags (27a8585)
- update links from S1M0N38 to coder (9681cf5)
0.11.0 (2025-10-03)
- add support for multiple seeds (a598e8e)
- default to cerebras provider if available (18b58ec)
- models: add deepseek v3.1 (561ac82)
- models: add google/gemini-2.5-flash (7807fbe)
- models: add grok-4-fast (85ebb9c)
- move seed into stats to support multi-seed runs (54775a7)
- remove blind select and cash_out phases from default strategy (bf8e0f2)
- add cards param to use_consumable in TOOLS.json (0faa085)
- add fallback for reasoning not available (fa530c5)
- add small sleeps in round eval and blind select (38fd254)
- include gamestate (SELECTING_HAND and SHOP) in GAMESTATE.md.jinja (7b2fcec)
- set seed to positive integer (b4c753a)
- typo and ensure benchmarks dir creation (cead758)
- add balatrobench images (8db3017)
- add first version of the docs (WIP) (74aa38f)
- add mkdocs.yml for generating documentation with mkdocs (72701ea)
- update analysis.md (903f4ab)
- update analysis.md (WIP) (b7ad0af)
0.10.0 (2025-09-22)
- add cli option to convert pngs to avif (3e8a655)
- change stats runs from int to list of str (7560acf)
- make use of balatrobot enums (8f7185b)
0.9.0 (2025-09-19)
- add avoiding the sell-buy trap in STRATEGY.md.jinja (8d163b1)
- add claude-sonnet-4 and gpt-5 to models (c2065bb)
- add current score and target score to gamestate (ae45b4d)
- add gemini-2.5-pro to models (e93e70d)
- add screenshot dir in data collection (ba8ee53)
- add shop, vouchers to GAMESTATE.md.jinja (760e5cb)
- add strategic discarding and suggestion about buy-sell (cfc2066)
- create details dir for each run in benchmark (30a4c8c)
- default actions for round_eval and blind_select (a273823)
- take screenshot on each request (4b25a07)
- computation of pooled variance in benchmark (ba2d522)
- consumables and tags representation in GAMESTATE.md.jinja (5930523)
- make the highlighted limit more explicit in default strategy (7dcaa59)
- regex for model name add dot (3aff39f)
- set "order of operations" section to h3 (172e90c)
- sort poker hands by order (2a5589a)
- update the BALATRO_CONSTANTS with proper values (e9a037f)
- use ante instead of computation of ante from round (72a0110)
- use state["game"]["round_resets"]["ante"] (5955982)
- whitespaces in GAMESTATE.md.jinja template (76067af)
0.8.2 (2025-09-09)
- add milliseconds to timestamp (8b271a8)
0.8.1 (2025-09-08)
- remove release process form CONTRIBUTING.md (3763cec)
0.8.0 (2025-09-06)
- add model-specific config loading (15d71d9)
- bot: improve error handling and LLM integration (310799c)
- move to openrouter (6a0f4fa)
- remove litellm dependency (28fb425)
- replace litellm config with models.yaml (446d981)
- use model config for model parameters (3576fd5)
- add strategy submission guidelines (03cc82d)
- update CLI and documentation for direct OpenRouter integration (00a061d)
0.7.0 (2025-09-03)
- add configurable port for balatrobot client (3a5cb23)
- add extra_headers to requests for App identification (65436cb)
- add extra_headers to requests for App identification (2ed4a1e)
- add fields provided by OpenRouter to stats.json (af97295)
- add fields provided by OpenRouter to stats.json (f4cd7fb)
- make balatro instance configurable in Makefile (1f7ad4b)
- parallelize runs over multiple ports (75286fa)
- parallelize runs with multiple ports (7169a24)
- stop game on 3 consecutive failed/error calls (3e25508)
- update benchmark with new openrouter fields (0c4c8a5)
- update default model to openai/gpt-oss-20b (c6b7cdc)
- update default model to openai/gpt-oss-20b (70bffc7)
- use openrouter in litellm config (6a42fe0)
- use openrouter in litellm config (85e9d22)
- update balatrobench command to new models (de5222d)
- update balatrobench command to new models (5042746)
- use invalid_responses instead of error_calls (9a09569)
- update help for balatrollm (6b4a515)
0.6.0 (2025-09-02)
- add last_response_is_invalid and last_tool_called_failed to LLMBot (7ae6d4a)
- make use of error messages in memory (1c8bc14)
- update data collection to include invalid responses (18e7341)
- update default MEMORY strategy with previous errors (c02a9b0)
- comment BotError to BalatroError (4b42402)
- remove reasoning effort from qwen-3-32b config (17265db)
- typos in data_collection.py (17e0542)
- use logger instead of logging in bot.py (d69ba00)
0.5.0 (2025-09-01)
- add benchmark command in Makefile (dcf0c56)
- add multi-run support and improve CLI argument handling (fd9219a)
- enhance bot architecture with comprehensive docstrings (02bfa5b)
- make base_dir configurable (c285c10)
- make number of runs configurable in Makefile (c47d373)
- report averaged stats in leaderboard (be5e381)
- update MEMORY.md.jinja with raw tool_calls (4493674)
- add game state 4 to completed (c6df03d)
- add balatrobench command to README and CLAUDE (6e1e9d5)
- add comprehensive documentation to benchmark system (e93dcf7)
- add comprehensive documentation to Config class (34308c2)
- add comprehensive documentation to data collection system (48e7cdf)
- add comprehensive documentation to strategy system (4e61cb7)
- update CLI documentation with new options (69cd0d0)
0.4.0 (2025-08-29)
- add new models and update old ones to new format (c99cb0a)
- add setup and teardown targets (e18a048)
- config: add metadata fields and simplify structure (aae7d16)
- remove support for env vars update flags (fe93936)
- update benchmark and data collection to new dir structure (1d21cbe)
- remove the footer form the readme (db5d4c2)
- update CLAUDE.md with help text from commands (f3b78af)
- update documentation to reflect architecture changes (181b4be)
- update README.md (da45438)
0.3.0 (2025-08-25)
- add additional fields to config.json for community metadata (68d011b)
- add audio support to balatro.sh (91e0d69)
- add benchmark subcommand to CLI (745b61f)
- add comprehensive benchmark analysis system (87f190b)
- enhance CLI with strategy system support (849a027)
- enhance configuration system with version validation and CLI overrides (5400f64)
- improve data collection and logging (c2685a9)
- re-work templates into four standard files (8410f0a)
- update Bot to use new template system (f1937a4)
- update core classes to use strategy system (296233c)
- update environment configuration for strategy system (94fa214)
- change the default template to
default(4a84cc8) - update config field references for consistency (ea4dc0b)
- add modern type hints guidelines to development documentation (374b365)
- update CLAUDE.md (756171b)
- update CLAUDE.md (487f8b6)
- update CLAUDE.md for strategy system (1157c7e)
- update README.md (ca19807)
0.2.0 (2025-08-21)
- add new tools and agentic capabilities (80e9d9a)
- add qwen3 to proxy config (b4a3da1)
- build: add game automation and enhanced testing support (9805c40)
- change default model to cerebras-qwen3-235b (f354789)
- core: refactor LLMBot with Config dataclass and improved architecture (f38d93d)
- deployment: add Balatro automation script (d5fdbcf)
- drastically simplify the templates (ef817f1)
- templates: add aggressive strategy template system (6c092b7)