pre-review fixes

CLAUDE.md

@@ -46,6 +46,7 @@ tests/                     # pytest, uses seeded_repo fixture from conftest.py
 
 ## Key patterns
 
+- **Guardrails are emergent, not authored.** Entities never reference each other by name. Each declares only its own intrinsic physical envelope (`requires`/`provides`/`excludes` for categorical fit, `range_min`/`range_max` for numeric floors/ceilings on shared keys like `mass`, `footprint`, `energy_density`). `ConstraintResolver` cross-checks these generically across every pair in a combo — a nuclear reactor is blocked from a bicycle because `mass range_min` collides with `mass range_max`, not because any code says "nuclear + bicycle = blocked." When adding physical realism to an entity, add a property to the entity — never a pairwise special case in `ConstraintResolver`. A category missing a floor it should have is a silent guardrail hole (blocks nothing, looks fine). See `LOGIC DOCS/002-guardrails-via-intrinsic-properties.md`.
 - **Repository is the only DB interface.** No raw SQL outside `repository.py`.
 - **Pipeline is combo-first**: each combo goes through all requested passes before the next combo starts. Progress is persisted per-combo (crash-safe, resumable).
 - **`pipeline_runs` table** tracks run lifecycle: pending → running → completed/failed/cancelled. The web route creates the record, then starts a background thread with its own `sqlite3.Connection`.

LOGIC DOCS/002-guardrails-via-intrinsic-properties.md

@@ -0,0 +1,37 @@
+# Guardrails via intrinsic properties, not special cases
+
+## The goal
+
+Full recombination across the existing catalog — any platform × any actuator × any energy source — should stay possible, while physically impossible pairings (a nuclear reactor on a bicycle, a solar sail in an atmosphere) get blocked automatically. Neither half is negotiable: too loose and the results are nonsense, too special-cased and the combinatorics stop being the point of the project.
+
+## The mechanism (already in place)
+
+Entities never reference each other. Each one only declares its own physical envelope:
+
+- `requires`/`provides`/`excludes` — categorical compatibility (`energy_form`, `medium`, `atmosphere`)
+- `range_min`/`range_max` — numeric floors and ceilings on a shared key (`mass`, `footprint`, `energy_density`, ...)
+
+`ConstraintResolver` (`src/physcom/engine/constraint_resolver.py`) then runs generic, entity-agnostic rules across every pair in a combo: requires-vs-excludes, mutual exclusion, range incompatibility, provides-vs-range deficit, unmet requirements. A combo gets blocked because two numbers or two category tags collided — never because code somewhere says "if X and Y, block." Blocking is emergent from the data, not authored per pair.
+
+This is what makes "nuclear reactor won't fit on a bicycle" work today: Nuclear Thermal Drive declares `mass range_min = 1500kg`, Light Personal Vehicle declares `mass range_max = 60kg`, and Rule 3 (`_check_range_incompatibility`) blocks the combo without either entity knowing the other exists.
+
+## Why this breaks silently
+
+The mechanism only blocks what the data describes. A category missing a `mass range_min` or `footprint range_min` it should physically have doesn't get blocked — not because the rule is wrong, but because nothing told the resolver there was a limit. This is an easy failure mode: the constraint system looks like it's working (it blocks the pairs someone thought to test) while quietly admitting nonsense pairs nobody happened to check.
+
+Two such gaps were found and fixed 2026-07 (see git history around this doc's commit):
+
+- Solar Sail declared its size under the key `surface_area` instead of `footprint` — a naming mismatch made it invisible to every cross-check, silently a no-op.
+- Nuclear Thermal Drive and Nuclear Fuel had a `mass range_min` but no `footprint range_min` — reactor mass was guarded, reactor size was not.
+
+Both were fixed by adding/renaming a `Dependency`, zero changes to `ConstraintResolver` itself. That's the intended shape of a fix here.
+
+## The standing question for review
+
+For every actuator and energy_storage entity: what is the **full set of intrinsic physical floors and ceilings** a real version of this technology imposes at any scale — structural mass, footprint, radiation shielding, minimum coolant/containment volume, thermal rejection surface area, etc. — and is each one expressed as a `range_min`/`range_max`/`provides`/`requires` dependency using a key that **already exists** in the vocabulary (`grep -oP 'Dependency\("[a-z]+", "\K[a-z_]+'` over `seed/transport_example.py` lists it), so it actually cross-checks against every other entity that shares that key?
+
+A missing floor is a silent guardrail hole, not a missing detail.
+
+## Non-goal
+
+Do not add pairwise special cases (`if actuator == "Nuclear Thermal Drive" and platform == "..."`). That defeats the architecture — the whole point is that the catalog recombines freely and physics falls out of shared, generic constraints. Any new realism belongs on the entity as a property, not in `ConstraintResolver` as a rule.

src/physcom/cli.py

@@ -227,40 +227,36 @@ def review(ctx, combination_id):
 
 @main.command()
 @click.argument("domain_name")
-@click.option("--format", "fmt", default="md", help="Export format (md)")
 @click.option("--top", "-n", default=20, type=int, help="Number of results to export")
 @click.option("--output", "-o", default=None, help="Output file path")
 @click.pass_context
-def export(ctx, domain_name, fmt, top, output):
+def export(ctx, domain_name, top, output):
-    """Export results to a report."""
+    """Export results to a Markdown report."""
     repo = _get_repo(ctx.obj["db"])
     top_results = repo.get_top_results(domain_name, limit=top)
     if not top_results:
         click.echo(f"No results for domain '{domain_name}'.")
         return
 
-    if fmt == "md":
+    lines = [f"# {domain_name} — Top {len(top_results)} Concepts\n"]
-        lines = [f"# {domain_name} — Top {len(top_results)} Concepts\n"]
+    for i, r in enumerate(top_results, 1):
-        for i, r in enumerate(top_results, 1):
+        combo = r["combination"]
-            combo = r["combination"]
+        entity_names = " + ".join(e.name for e in combo.entities)
-            entity_names = " + ".join(e.name for e in combo.entities)
+        lines.append(f"## {i}. {entity_names} (score: {r['composite_score']:.4f})")
-            lines.append(f"## {i}. {entity_names} (score: {r['composite_score']:.4f})")
+        if r["novelty_flag"]:
-            if r["novelty_flag"]:
+            lines.append(f"**Novelty:** {r['novelty_flag']}")
-                lines.append(f"**Novelty:** {r['novelty_flag']}")
+        if r["llm_review"]:
-            if r["llm_review"]:
+            lines.append(f"\n{r['llm_review']}")
-                lines.append(f"\n{r['llm_review']}")
+        if r["human_notes"]:
-            if r["human_notes"]:
+            lines.append(f"\n*Notes:* {r['human_notes']}")
-                lines.append(f"\n*Notes:* {r['human_notes']}")
+        lines.append("")
-            lines.append("")
 
-        content = "\n".join(lines)
+    content = "\n".join(lines)
-        if output:
+    if output:
-            Path(output).write_text(content)
+        Path(output).write_text(content)
-            click.echo(f"Exported to {output}")
+        click.echo(f"Exported to {output}")
-        else:
-            click.echo(content)
     else:
-        click.echo(f"Unsupported format: {fmt}")
+        click.echo(content)
 
 
 @main.group()

src/physcom/db/repository.py

@@ -268,8 +268,8 @@ class Repository:
             by_key.setdefault(r["key"], []).append(r["value"])
         return [DomainConstraint(key=k, allowed_values=v) for k, v in by_key.items()]
 
-    def get_domain(self, name: str) -> Domain | None:
+    def _load_domain(self, where: str, param: str | int) -> Domain | None:
-        row = self.conn.execute("SELECT * FROM domains WHERE name = ?", (name,)).fetchone()
+        row = self.conn.execute(f"SELECT * FROM domains WHERE {where} = ?", (param,)).fetchone()
         if not row:
             return None
         weights = self.conn.execute(
@@ -296,37 +296,15 @@ class Repository:
             constraints=self._load_domain_constraints(row["id"]),
         )
 
+    def get_domain(self, name: str) -> Domain | None:
+        return self._load_domain("name", name)
+
     def list_domains(self) -> list[Domain]:
         rows = self.conn.execute("SELECT name FROM domains ORDER BY name").fetchall()
         return [self.get_domain(r["name"]) for r in rows]
 
     def get_domain_by_id(self, domain_id: int) -> Domain | None:
-        row = self.conn.execute("SELECT * FROM domains WHERE id = ?", (domain_id,)).fetchone()
+        return self._load_domain("id", domain_id)
-        if not row:
-            return None
-        weights = self.conn.execute(
-            """SELECT m.name, m.unit, dmw.weight, dmw.norm_min, dmw.norm_max,
-                      dmw.metric_id, dmw.lower_is_better
-               FROM domain_metric_weights dmw
-               JOIN metrics m ON dmw.metric_id = m.id
-               WHERE dmw.domain_id = ?""",
-            (row["id"],),
-        ).fetchall()
-        return Domain(
-            id=row["id"],
-            name=row["name"],
-            description=row["description"] or "",
-            metric_bounds=[
-                MetricBound(
-                    metric_name=w["name"], weight=w["weight"],
-                    norm_min=w["norm_min"], norm_max=w["norm_max"],
-                    metric_id=w["metric_id"], unit=w["unit"] or "",
-                    lower_is_better=bool(w["lower_is_better"]),
-                )
-                for w in weights
-            ],
-            constraints=self._load_domain_constraints(row["id"]),
-        )
 
     def update_domain(self, domain_id: int, name: str, description: str) -> None:
         self.conn.execute(

src/physcom/llm/providers/ollama.py

@@ -0,0 +1,73 @@
+"""Ollama LLM provider — local models via the Ollama HTTP API."""
+
+from __future__ import annotations
+
+import json
+import re
+import urllib.error
+import urllib.request
+
+from physcom.llm.base import LLMProvider
+from physcom.llm.prompts import PHYSICS_ESTIMATION_PROMPT, PLAUSIBILITY_REVIEW_PROMPT
+
+
+class OllamaLLMProvider(LLMProvider):
+    """LLM provider backed by a local Ollama server (see `ollama serve`)."""
+
+    def __init__(self, model: str = "qwen2.5:7b", host: str = "http://localhost:11434") -> None:
+        self._model = model
+        self._host = host.rstrip("/")
+
+    def estimate_physics(
+        self, combination_description: str, metrics: list[str]
+    ) -> dict[str, float]:
+        prompt = PHYSICS_ESTIMATION_PROMPT.format(
+            description=combination_description,
+            metrics=", ".join(metrics),
+        )
+        text = self._generate(prompt, json_mode=True)
+        return self._parse_json(text, metrics)
+
+    def review_plausibility(
+        self, combination_description: str, scores: dict[str, float]
+    ) -> tuple[str, bool]:
+        scores_str = "\n".join(f"- {k}: {v:.3f}" for k, v in scores.items())
+        prompt = PLAUSIBILITY_REVIEW_PROMPT.format(
+            description=combination_description,
+            scores=scores_str,
+        )
+        text = self._generate(prompt, json_mode=False).strip()
+        return (text, self._parse_verdict(text))
+
+    def _generate(self, prompt: str, json_mode: bool) -> str:
+        payload = {"model": self._model, "prompt": prompt, "stream": False}
+        if json_mode:
+            payload["format"] = "json"
+        req = urllib.request.Request(
+            f"{self._host}/api/generate",
+            data=json.dumps(payload).encode("utf-8"),
+            headers={"Content-Type": "application/json"},
+        )
+        try:
+            with urllib.request.urlopen(req, timeout=120) as resp:
+                return json.loads(resp.read())["response"]
+        except urllib.error.URLError as exc:
+            raise ConnectionError(
+                f"Could not reach Ollama at {self._host} (is `ollama serve` running?)"
+            ) from exc
+
+    def _parse_verdict(self, text: str) -> bool:
+        """Extract VERDICT: PLAUSIBLE/IMPLAUSIBLE from response; default to True."""
+        m = re.search(r"VERDICT:\s*(PLAUSIBLE|IMPLAUSIBLE)", text, re.IGNORECASE)
+        if m:
+            return m.group(1).upper() == "PLAUSIBLE"
+        return True
+
+    def _parse_json(self, text: str, metrics: list[str]) -> dict[str, float]:
+        """Strip markdown fences and parse JSON; fall back to 0.5 per metric on error."""
+        text = re.sub(r"```(?:json)?\s*", "", text).strip().rstrip("`").strip()
+        try:
+            data = json.loads(text)
+            return {k: float(v) for k, v in data.items() if k in metrics}
+        except (json.JSONDecodeError, ValueError, TypeError):
+            return {m: 0.5 for m in metrics}

src/physcom/llm/registry.py

@@ -10,9 +10,11 @@ from physcom.llm.base import LLMProvider
 def build_llm_provider() -> LLMProvider | None:
     """Return an LLMProvider based on env vars, or None if not configured.
 
-    LLM_PROVIDER   — provider name ('gemini'; more can be added)
+    LLM_PROVIDER   — provider name ('gemini', 'ollama'; more can be added)
     GEMINI_API_KEY — required when LLM_PROVIDER=gemini
     GEMINI_MODEL   — optional Gemini model name (default: gemini-2.0-flash)
+    OLLAMA_MODEL   — optional Ollama model name (default: qwen2.5:7b)
+    OLLAMA_HOST    — optional Ollama server URL (default: http://localhost:11434)
     """
     provider = os.environ.get("LLM_PROVIDER", "").lower().strip()
 
@@ -27,4 +29,10 @@ def build_llm_provider() -> LLMProvider | None:
         from physcom.llm.providers.gemini import GeminiLLMProvider
         return GeminiLLMProvider(api_key=api_key, model=model)
 
-    raise ValueError(f"Unknown LLM_PROVIDER: {provider!r}. Supported: gemini")
+    if provider == "ollama":
+        model = os.environ.get("OLLAMA_MODEL", "qwen2.5:7b")
+        host = os.environ.get("OLLAMA_HOST", "http://localhost:11434")
+        from physcom.llm.providers.ollama import OllamaLLMProvider
+        return OllamaLLMProvider(model=model, host=host)
+
+    raise ValueError(f"Unknown LLM_PROVIDER: {provider!r}. Supported: gemini, ollama")

src/physcom/seed/transport_example.py

@@ -338,7 +338,7 @@ RENEWABLE_ACTUATORS: list[Entity] = [
             Dependency("energy", "energy_form", "radiation_pressure", None, "requires"),
             Dependency("environment", "atmosphere", "vacuum_or_thin", None, "requires"),
             Dependency("environment", "star_proximity", "true", None, "requires"),
-            Dependency("physical", "surface_area", "100", "m²", "range_min"),
+            Dependency("physical", "footprint", "100", "m²", "range_min"),
             Dependency("force", "thrust_profile", "continuous_low", None, "provides"),
             Dependency("force", "power_density", "0.01", "W/kg", "provides"),
             Dependency("environment", "medium", "space", None, "requires"),
@@ -393,6 +393,7 @@ ROCKET_ACTUATORS: list[Entity] = [
         dependencies=[
             Dependency("energy", "energy_form", "nuclear_thermal", None, "requires"),
             Dependency("physical", "mass", "1500", "kg", "range_min"),
+            Dependency("physical", "footprint", "20", "m²", "range_min"),
             Dependency("force", "thrust_profile", "extreme_continuous", None, "provides"),
             Dependency("force", "power_density", "50", "W/kg", "provides"),
             Dependency("material", "radiation_shielding", "true", None, "requires"),
@@ -644,6 +645,7 @@ PROPELLANT_STORAGE: list[Entity] = [
             Dependency("energy", "energy_form", "nuclear_thermal", None, "provides"),
             Dependency("infrastructure", "fuel_infrastructure", "nuclear_fuel", None, "requires"),
             Dependency("physical", "mass", "500", "kg", "range_min"),
+            Dependency("physical", "footprint", "5", "m²", "range_min"),
             Dependency("physical", "energy_density", "1800000000", "J/kg", "provides"),
             Dependency("material", "radiation_shielding", "true", None, "requires"),
         ],

src/physcom_web/app.py

@@ -2,7 +2,6 @@
 
 from __future__ import annotations
 
-import math
 import os
 import secrets
 from pathlib import Path
@@ -55,53 +54,10 @@ def close_db(exc: BaseException | None = None) -> None:
         repo.conn.close()
 
 
-_SI_PREFIXES = [
-    (1e12, "T"),
-    (1e9, "G"),
-    (1e6, "M"),
-    (1e3, "k"),
-]
-
-
-def _si_format(value: object) -> str:
-    """Format a number with SI prefixes for readability.
-
-    Handles string inputs (like dep.value) by trying float conversion first.
-    Non-numeric values are returned as-is.
-    """
-    if isinstance(value, str):
-        try:
-            num = float(value)
-        except (ValueError, TypeError):
-            return value
-    elif isinstance(value, (int, float)):
-        num = float(value)
-    else:
-        return str(value)
-
-    if math.isnan(num) or math.isinf(num):
-        return str(value)
-
-    abs_num = abs(num)
-    if abs_num < 1000:
-        # Small numbers: drop trailing zeros, cap at 4 significant figures
-        if num == int(num) and abs_num < 100:
-            return str(int(num))
-        return f"{num:.4g}"
-
-    for threshold, prefix in _SI_PREFIXES:
-        if abs_num >= threshold:
-            scaled = num / threshold
-            return f"{scaled:.4g}{prefix}"
-
-    return f"{num:.4g}"
-
-
 def create_app() -> Flask:
     app = Flask(__name__)
     app.secret_key = _load_or_generate_secret_key()
 
-    app.jinja_env.filters["si"] = _si_format
     app.jinja_env.filters["qty"] = format_quantity
 
     app.teardown_appcontext(close_db)

src/physcom_web/routes/entities.py

@@ -59,12 +59,6 @@ def entity_detail(entity_id: int):
     return render_template("entities/detail.html", entity=entity)
 
 
-@bp.route("/<int:entity_id>/edit")
-def entity_edit(entity_id: int):
-    """Legacy route — redirect to detail page."""
-    return redirect(url_for("entities.entity_detail", entity_id=entity_id))
-
-
 @bp.route("/<int:entity_id>/delete", methods=["POST"])
 def entity_delete(entity_id: int):
     repo = get_repo()

src/physcom_web/templates/entities/list.html

@@ -31,7 +31,7 @@
           <td>{{ e.description }}</td>
           <td>{{ e.dependencies|length }}</td>
           <td>
-            <a href="{{ url_for('entities.entity_edit', entity_id=e.id) }}" class="btn btn-sm">Edit</a>
+            <a href="{{ url_for('entities.entity_detail', entity_id=e.id) }}" class="btn btn-sm">Edit</a>
           </td>
         </tr>
         {% endfor %}

tests/test_ollama_provider.py

@@ -0,0 +1,43 @@
+"""Tests for the Ollama provider's parsing logic and registry wiring."""
+
+from __future__ import annotations
+
+import pytest
+
+from physcom.llm.providers.ollama import OllamaLLMProvider
+
+
+@pytest.fixture
+def provider():
+    return OllamaLLMProvider()
+
+
+def test_parse_json_strips_fences(provider):
+    text = '```json\n{"power_density": 500.0, "safety": 0.7}\n```'
+    result = provider._parse_json(text, ["power_density", "safety"])
+    assert result == {"power_density": 500.0, "safety": 0.7}
+
+
+def test_parse_json_falls_back_on_invalid(provider):
+    result = provider._parse_json("not json", ["power_density", "safety"])
+    assert result == {"power_density": 0.5, "safety": 0.5}
+
+
+def test_parse_verdict_plausible(provider):
+    assert provider._parse_verdict("blah blah\nVERDICT: PLAUSIBLE") is True
+
+
+def test_parse_verdict_implausible(provider):
+    assert provider._parse_verdict("blah blah\nVERDICT: IMPLAUSIBLE") is False
+
+
+def test_registry_builds_ollama_provider(monkeypatch):
+    from physcom.llm.registry import build_llm_provider
+
+    monkeypatch.setenv("LLM_PROVIDER", "ollama")
+    monkeypatch.setenv("OLLAMA_MODEL", "phi4:14b")
+    monkeypatch.setenv("OLLAMA_HOST", "http://example:1234")
+    provider = build_llm_provider()
+    assert isinstance(provider, OllamaLLMProvider)
+    assert provider._model == "phi4:14b"
+    assert provider._host == "http://example:1234"