6afdb4b148
Lucidia Core - AI reasoning engines for specialized domains: - Physicist (867 lines) - energy modeling, force calculations - Mathematician (760 lines) - symbolic computation, proofs - Geologist (654 lines) - terrain modeling, stratigraphy - Engineer (599 lines) - structural analysis, optimization - Painter (583 lines) - visual generation, graphics - Chemist (569 lines) - molecular analysis, reactions - Analyst (505 lines) - pattern recognition, insights - Plus: architect, researcher, mediator, speaker, poet, navigator Features: - FastAPI wrapper with REST endpoints for each agent - CLI with `lucidia list`, `lucidia run`, `lucidia api` - Codex YAML configurations for agent personalities - Quantum engine extensions 12,512 lines of Python across 91 files. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
95 lines
3.4 KiB
Python
95 lines
3.4 KiB
Python
"""Generator interfaces for Lucidia's reasoning duet."""
|
|
|
|
from __future__ import annotations
|
|
|
|
from dataclasses import dataclass, field
|
|
from typing import Any, Callable, Dict, List, Optional, Protocol
|
|
|
|
|
|
@dataclass
|
|
class ProposeInput:
|
|
"""Minimal information required to form a proposal."""
|
|
|
|
goal: str
|
|
context: Dict[str, Any]
|
|
constraints: Optional[List[str]] = None
|
|
|
|
|
|
@dataclass
|
|
class Proposal:
|
|
"""Structured generator response."""
|
|
|
|
summary: str
|
|
rationale: List[str]
|
|
plan: List[str]
|
|
raw: Dict[str, Any] = field(default_factory=dict)
|
|
|
|
def to_payload(self) -> Dict[str, Any]:
|
|
"""Convert to a serialisable payload."""
|
|
|
|
return {
|
|
"summary": self.summary,
|
|
"rationale": list(self.rationale),
|
|
"plan": list(self.plan),
|
|
"raw": dict(self.raw),
|
|
}
|
|
|
|
|
|
class GeneratorBackend(Protocol):
|
|
"""Callable signature for lightweight local LLMs."""
|
|
|
|
def __call__(self, prompt: str) -> Dict[str, Any]: # pragma: no cover - protocol definition
|
|
...
|
|
|
|
|
|
class LocalGenerator:
|
|
"""Wraps a lightweight local LLM backend with prompt scaffolding."""
|
|
|
|
def __init__(self, model_name: str, backend: GeneratorBackend, prompt_builder: Callable[[ProposeInput], str]) -> None:
|
|
self.model_name = model_name
|
|
self._backend = backend
|
|
self._prompt_builder = prompt_builder
|
|
|
|
def build_prompt(self, payload: ProposeInput) -> str:
|
|
"""Build the structured system prompt."""
|
|
|
|
return self._prompt_builder(payload)
|
|
|
|
def propose(self, payload: ProposeInput) -> Proposal:
|
|
"""Invoke the backend and normalise output."""
|
|
|
|
prompt = self.build_prompt(payload)
|
|
response = self._backend(prompt)
|
|
summary = response.get("summary") or response.get("SUMMARY")
|
|
rationale = response.get("rationale") or response.get("RATIONALE")
|
|
plan = response.get("plan") or response.get("PLAN")
|
|
|
|
if not isinstance(summary, str):
|
|
raise ValueError("Generator response must include a 'summary' string.")
|
|
if not isinstance(rationale, list) or not all(isinstance(item, str) for item in rationale):
|
|
raise ValueError("Generator response must include a 'rationale' list of strings.")
|
|
if not isinstance(plan, list) or not all(isinstance(item, str) for item in plan):
|
|
raise ValueError("Generator response must include a 'plan' list of strings.")
|
|
|
|
raw = {key: value for key, value in response.items() if key not in {"summary", "SUMMARY", "rationale", "RATIONALE", "plan", "PLAN"}}
|
|
|
|
return Proposal(summary=summary.strip(), rationale=[item.strip() for item in rationale], plan=[item.strip() for item in plan], raw=raw)
|
|
|
|
|
|
def default_prompt_builder(payload: ProposeInput) -> str:
|
|
"""Default prompt template aligned with duet requirements."""
|
|
|
|
constraints = "\n".join(f"- {c}" for c in (payload.constraints or []))
|
|
context_repr = "\n".join(f"{key}: {value}" for key, value in payload.context.items())
|
|
return (
|
|
"SYSTEM: You propose; Codex Ψ′ will validate. No filler. Provide:\n"
|
|
"1) SUMMARY (≤120 words), 2) RATIONALE (bullet steps), 3) PLAN (3 actions).\n"
|
|
"CONSTRAINTS:\n"
|
|
f"{constraints}\n"
|
|
"TASK: {goal}\n"
|
|
"CONTEXT:\n{context}\n"
|
|
).format(goal=payload.goal, context=context_repr)
|
|
|
|
|
|
__all__ = ["Proposal", "ProposeInput", "LocalGenerator", "default_prompt_builder"]
|