Repository navigation
Review fixes - #4
Merged
Merged
Conversation
- TextWorld is now the single source of truth for each agent's hunger,
mood, goal, position and inventory; hunger rises in world.update().
- Add a VERBS registry (+ aliases) and validate_action(), so every action
must use a known verb and a target the agent can actually reach.
- Scenario loader reads objects declared inside rooms, derives affordances
from object type, supports locked doors, per-scenario hunger rate and
toggles for random events / neighbor visits.
- Items in open containers can be taken directly ("take key").
- Restore idle-object perception and surface the SensoryCortex qualia.
- Every scenario now declares a scripted SOLUTION and FAILURE, and
tests/test_scenarios.py proves each is winnable and losable.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- CognitiveLoop.step() runs one OODA cycle; run_loop() and benchmark.py both call it, so benchmarks exercise the real cognition. - agent_status is a read-only view of the world's agent; a fresh world is seeded from config, a scenario world keeps its own state. - Apply the subconscious emotional shift once per cycle (it was applied twice, doubling every level_delta). - Drop ungrounded impulses before imagining; replace an ungrounded final decision with the strongest grounded impulse. - Goals: adopt a proposed goal only when none is active; support goal_status completed/abandoned. - Theory of Mind runs when another agent speaks or at most every TOM_INTERVAL cycles, using what they actually said. - Default AGENT_GOAL is now empty so Adam proposes his own goals. - Fix /get_state current_goal, which read a non-existent attribute. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Replace instructor + openai with ollama.chat(format=<json schema>), so
replies always parse; the verb field is an enum built from VERBS.
- Re-ask the model with the validation error when a reply doesn't fit.
- Collapse the six endpoint bodies into one handler; fallbacks are marked
with "psyche_fallback": true.
- Prompts: generate the verb toolbox from the registry, show reachable
objects/items/exits/doors, give reflection the world state and a
goal_status field.
- Fix failed-action counting ("open the" -> "open fridge").
- Pass adamsec prompt-injection text through to the prompts so the
harness actually tests something.
- Refresh requirements pins to the installed versions.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
SENTENCE_MODEL now defaults to all-MiniLM-L6-v2. If memory cannot start (no model, load failure, Chroma/Pinecone unavailable) MemoryStore logs a single LONG-TERM MEMORY DISABLED warning with the reason. MEMORY_BACKEND=none turns memory off quietly. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Runs on Python 3.10 and 3.13 with CPU-only torch to keep installs small. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- README/MANUAL: Chroma default memory, Ollama structured outputs, verb registry workflow, scenario format with SOLUTION/FAILURE, goal lifecycle. - .env.example: psyche port 5001 (matches the server), llama3 default, empty AGENT_GOAL, memory backend options. - Mark adamsec as experimental; move the QuTiP demo to experiments/. - Stop tracking .DS_Store and .psyche_ui.json. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Small models fill the field even when Adam is alone, which made the log report imaginary Theory of Mind. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Merged
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.