Consolidate clean base + bugbot fixes #8

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open

JackSwitzer wants to merge 30 commits into main from consolidation/clean-base-2026-02-03

CLAUDE.md

            
                      Original file line number
                      Diff line number
                      Diff line change
                  
    @@ -7,7 +7,7 @@ Build a mod/bot that wins Slay the Spire (Watcher only, A20, >96% winrate) using
  
    ```

    packages/engine/     # Pure Python game engine (source of truth)

    packages/parity/     # Seed catalog + parity verification tools

    tests/               # 4100+ tests (pytest)

    tests/               # 4500+ tests (pytest)

    mod/                 # Java EVTracker mod

    decompiled/          # Java source reference

    docs/vault/          # Game mechanics ground truth

    @@ -16,26 +16,42 @@ docs/                # ARCHITECTURE.md
  
    ## Testing

    ```bash

    uv run pytest tests/ -q              # Run all 4100+ tests

    uv run pytest tests/ -q              # Run all 4500+ tests

    uv run pytest tests/test_parity.py   # Parity verification

    uv run pytest tests/ --cov=packages/engine  # Coverage (~68%)

    ```

    ## Java Parity Status (Verified 2026-02-03)

    | Domain | Parity | Status |

    |--------|--------|--------|

    | **RNG System** | 100% | ✅ PRODUCTION READY |

    | **Damage/Block Calc** | 100% | ✅ Exact match |

    | **Enemies** | 100% | ✅ All 66 verified |

    | **Stances** | 100% | ✅ Divinity timing fixed |

    | **Cards** | 97% | ✅ 6 cards fixed |

    | **Powers** | 84% | ⚠️ Most triggers working |

    | **Events** | 95% | ⚠️ Minor issues |

    | **Potions (Data)** | 100% | ✅ All 42 correct |

    | **Potions (Effects)** | 40% | ⚠️ 25 missing effects |

    | **Relics (Active)** | 65% | ⚠️ 117/180 triggers implemented |

    | **Relics (Passive)** | 15% | ✅ Data-driven (27/180) |

    ## Java Parity Status (Verified 2026-02-04)

    ### Core Mechanics (100% Parity)

    | Domain | Status | Notes |

    |--------|--------|-------|

    | **RNG System** | ✅ 100% | All 13 streams verified, production ready |

    | **Damage Calculation** | ✅ 100% | Vuln 1.5x, Weak 0.75x, floor operations exact |

    | **Block Calculation** | ✅ 100% | Dex before Frail, floor operations exact |

    | **Stances** | ✅ 100% | Wrath/Calm/Divinity/Neutral verified |

    | **Enemies** | ✅ 100% | All 66 enemies verified |

    | **Map Generation** | ✅ 100% | Java quirks included |

    | **Shop** | ✅ 100% | Prices and pools match |

    | **Card Rewards** | ✅ 100% | HashMap order + pity timer |

    | **Card Data** | ✅ 100% | All characters verified |

    | **Potions (Data)** | ✅ 100% | All 42 potions correct |

    ### Partial / Missing (Implementation Gaps)

    - **Power triggers**: 30/94 implemented (64 missing).

    - **Relics**: 44 missing all + rest-site/pickup/chest/bottled gaps (see skipped tests).

    - **Events**: 17/50 choice generators implemented; 2 handlers missing.

    - **Potions (effects)**: discovery/selection and several effects partial.

    - **Rewards**: JSON action layer implemented; fidelity still depends on relic/potion/event gaps.

    ### Missing Features (138 Skipped Tests by markers)

    | Category | Count | Priority | Description |

    |----------|-------|----------|-------------|

    | Rest Site Relics | 36 | HIGH | Dream Catcher, Regal Pillow, Girya, Peace Pipe, Shovel |

    | Relic Pickup Effects | 34 | HIGH | War Paint, Whetstone, Astrolabe, Calling Bell, etc. |

    | Chest Relic Acquisition | 30 | HIGH | Tiny Chest, Matryoshka, Black Star, Cursed Key |

    | Bottled Relics | 20 | MED | Bottled Flame/Lightning/Tornado innate hands |

    | Out-of-Combat Triggers | 13 | MED | Shop relics, Ectoplasm tracking |

    See test files and `docs/ARCHITECTURE.md` for implementation details.

    @@ -45,8 +61,8 @@ from packages.engine import GameRunner, GamePhase
  
    runner = GameRunner(seed="SEED", ascension=20)

    while not runner.game_over:

        actions = runner.get_available_actions()  # Decision point

        runner.take_action(actions[0])            # Execute action

        actions = runner.get_available_action_dicts()  # JSON actions

        runner.take_action_dict(actions[0])            # Execute action

        # runner.run_state = full observable state

        # runner.phase = current GamePhase

    ```

    @@ -259,6 +275,13 @@ Tier 2: InnerPeace, CutThroughFate, WheelKick, Conclude
  
    - Divinity enter: +3

    - Deva Form: +1/turn stacking

    ### Scry Mechanics

    - Scry X: Look at top X cards of draw pile, choose which to DISCARD (rest stay on top)

    - Golden Eye relic: +2 to ALL scry amounts

    - Melange relic: Scry 3 on shuffle

    - Nirvana power: Gain block once per scry ACTION (not per card)

    - Agent decides which cards to discard via `SelectScryDiscard` action

    ## RNG Prediction System (✅ Verified 100% Java Parity)

    ### Documentation

docs/ARCHITECTURE.md

-Original file line number
+Diff line change
@@ Expand Up / @@ -42,7 +42,7 @@ SlayTheSpireRL/ @@
     │   └── combat_engine.py    # Turn-based combat engine
     ├── packages/parity/         # Seed parity verification tools
     │   └── comparison/         # State comparators, RNG trackers, game simulator
-    ├── tests/                   # 2480+ pytest tests
+    ├── tests/                   # 4500+ pytest tests
     ├── mod/                     # Java EVTracker mod
     ├── docs/                    # Documentation
     │   └── vault/              # Verified game mechanics ground truth
@@ Expand Down @@

docs/GAME_STATE_AUDIT.md

            
                      Original file line number
                      Diff line number
                      Diff line change
                  
    @@ -50,11 +50,11 @@ Note:
  
    ### Phase/Flow State (`GameRunner`)

    Implemented:

    - Full phase machine: map navigation, combat, rewards, events, shops, rest, treasure, boss rewards, Neow.

    - Action types for all phases for RL integration.

    - JSON action layer + observation API for RL integration.

    Present but partial:

    - `GameRunner` supports Watcher/Ironclad/Silent/Defect run factories.

    - Some phase-specific actions are defined but their handler types are still stubs (see Reward actions).

    - Some phase outcomes still depend on missing relic/potion/event behaviors.

    ### Room/Modal State

    Implemented dataclasses:

    @@ -63,30 +63,26 @@ Implemented dataclasses:
  
    - Reward types (`CombatRewards`, `CardReward`, etc.).

    Present but partial:

    - Reward action dataclasses are `pass` (claim/skip/proceed actions are not fully implemented).

    - Reward action processing is implemented, but fidelity depends on missing relic/potion/event hooks.

    - Event choice generators and handler coverage are incomplete; ID normalization now handled via alias mapping.

    ## What’s Missing / Needs Work

    ### Core State Gaps

    - **Orb system**: channel/evoke, slots, Focus/Lock‑On, orb-specific state in `CombatState`.

    - **Reward action processing**: Claim/skip actions in `RewardHandler` (currently stubs).

    - **Event normalization**: alias mapping added; full content/handler unification and missing choice generators remain.

    ### Combat/State Hooks

    - Combat‑time relic hooks are incomplete (`Snecko Eye`, Ice Cream, etc.).

    - Potion effect behaviors are still simplified (Discovery choices, Distilled Chaos auto-play, Entropic Brew RNG parity, Smoke Bomb restrictions).

    - Power triggers missing across multiple hooks (see `docs/work_units/powers.md`).

    - Power triggers missing across multiple hooks (see `docs/work_units/granular-powers.md`).

    ### Data/Pool Consistency

    - Legacy Java IDs remain canonical, but modern names are now supported via alias mapping (Rushdown → `Adaptation`, Foresight → `Wireheading`, Wraith Form → `Wraith Form v2`).

    ### Tests and Known Failures (latest run)

    Full test run: `uv run pytest tests/ -ra` (2026-02-04)

    - Collected 3950 tests; run aborted during collection with 3 import errors:

      - `tests/test_ascension.py`: import error for `WATCHER_BASE_GOLD` (removed in favor of `BASE_STARTING_GOLD`).

      - `tests/test_coverage_boost.py`: import error for `EventHandler` from `handlers.rooms` (moved to `handlers.event_handler`).

      - `tests/test_handlers.py`: import error for `EventHandler` from `handlers.rooms` (moved to `handlers.event_handler`).

    ### Tests and Known Failures (latest known)

    - Skip markers indicate **~138 skipped tests**, mostly relic pickup/rest-site/chest mechanics.

    - Targeted readiness tests for JSON actions + observations now pass.

    ### Watcher RL Readiness (current max)

    Safe (high parity):

    @@ -102,9 +98,8 @@ Cautious (partial parity; usable with constraints):
  
    - Events: missing choice generators default to leave; pool coverage incomplete.

    Risky (training fidelity compromised):

    - Reward action processing still incomplete (claim/skip action classes are stubs).

    - Defect orb system missing (avoid Prismatic Shard/cross-class reliance).

    - Known test import errors indicate test suite needs refactor updates before using tests as regression guard.

    - Some relic/potion/event behavior gaps can bias learning if not constrained.

    Practical constraints to “push max” now:

    - Prefer Watcher-only runs; avoid Prismatic Shard and cross-class effects.

    @@ -121,7 +116,6 @@ Implemented:
  
    Missing / Partial:

    - Orb system and Defect state.

    - Reward action processing.

    - Event ID normalization and missing choice generators.

    - Combat relic/potion hooks and power trigger coverage.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Consolidate clean base + bugbot fixes #8

Uh oh!

Diff view

Diff view

There are no files selected for viewing

Uh oh!

Consolidate clean base + bugbot fixes #8

Are you sure you want to change the base?

Uh oh!

Consolidate clean base + bugbot fixes #8

Uh oh!

Uh oh!

Diff view

Diff view

There are no files selected for viewing

Uh oh!