fix: CodeRabbit review fixes for Layer 3 (missed by the #2 merge)

README.md

@@ -94,7 +94,7 @@ tool = "get_transactions"
 args = { account_id = "acct-001" }
 ```
 
-`covenant snapshot` runs each probe and stores its response **fingerprint** — the type shape of what actually came back, never the values, which legitimately change — in the lock. `covenant check` re-runs the probes and classifies shape drift with the same severity model, at location `behavior`:
+`covenant snapshot` runs each probe and stores its response **fingerprint** — the type shape of what actually came back — in the lock, alongside one **sample** response so the optional `--judge` pass has a baseline to compare meaning against. Shape drift is judged on the fingerprint alone; values legitimately change between runs. `covenant check` re-runs the probes and classifies shape drift with the same severity model, at location `behavior`:
 
 ```bash
 $ COVENANT_BEHAVIOR_DRIFT=1 covenant check   # response body renames a field; schema untouched

covenant/cli.py

@@ -62,14 +62,16 @@ def _check_probes(
     from .judge import judge_probe  # the [judge] extra is optional; import on use
 
     descriptions = {t["name"]: t.get("description") for t in base_tools}
-    shape_drifted = {c.tool for c in changes}
+    # A probe is identified by tool + args, so judge each live probe on its own
+    # merits: skip only the probes that themselves errored or shape-drifted, never
+    # a clean probe that happens to share a tool name with a drifted sibling.
     for r in live:
-        if r["is_error"] or r["tool"] in shape_drifted:
+        base = base_by[probe_key(r["tool"], r["args"])]  # every live key is baselined (see above)
+        if r["is_error"] or diff_probes([base], [r]):
             continue  # errors and shape drift are already reported; judge only clean shapes
-        sample = base_by[probe_key(r["tool"], r["args"])].get("sample")
         verdict = judge_probe(
             r["tool"], descriptions.get(r["tool"]), r["args"],
-            sample, r["response"], model=cfg.judge_model,
+            base.get("sample"), r["response"], model=cfg.judge_model,
         )
         if verdict.drift:
             changes.append(Change(

covenant/config.py

@@ -77,7 +77,10 @@ def load_config(path: str | Path = "covenant.toml", server_override: str | None
     if not command and not url:
         raise ConfigError("no server configured: set [server].command or .url, or pass --server")
 
-    judge_model = (data.get("judge") or {}).get("model")
+    judge = data.get("judge") or {}
+    if not isinstance(judge, dict):
+        raise ConfigError("[judge] must be a table")
+    judge_model = judge.get("model")
     if judge_model is not None and not isinstance(judge_model, str):
         raise ConfigError("[judge].model must be a string")
 

covenant/fingerprint.py

@@ -15,6 +15,7 @@
 import json
 
 from ._types import JsonDict
+from .errors import CovenantError
 
 
 def fingerprint(value: object) -> JsonDict:
@@ -33,7 +34,7 @@ def fingerprint(value: object) -> JsonDict:
         if shapes and all(s == shapes[0] for s in shapes):
             return {"type": "array", "items": shapes[0]}
         return {"type": "array"}
-    raise ValueError(f"cannot fingerprint non-JSON value of type {type(value).__name__}")
+    raise CovenantError(f"cannot fingerprint non-JSON value of type {type(value).__name__}")
 
 
 def probe_key(tool: str, args: JsonDict | None) -> str:

covenant/introspect.py

@@ -92,7 +92,10 @@ async def _run_probes(config: Config, probes: list[Probe]) -> list[JsonDict]:
     records: list[JsonDict] = []
     async with _session(config) as session:
         for probe in probes:
-            result = await session.call_tool(probe.tool, probe.args)
+            try:
+                result = await session.call_tool(probe.tool, probe.args)
+            except Exception as e:  # noqa: BLE001 - name the probe so a config typo isn't a "connection" error
+                raise CovenantError(f"probe {probe.tool} failed: {e}") from e
             response, is_error, error = _resolve_result(result)
             records.append({
                 "tool": probe.tool,

covenant/judge/__init__.py

@@ -110,6 +110,9 @@ def judge_probe(
         raw = raw.strip("`").removeprefix("json").strip()
     try:
         data = json.loads(raw)
-        return Verdict(drift=bool(data["drift"]), reason=str(data["reason"]))
+        drift = data["drift"]
+        if not isinstance(drift, bool):
+            raise TypeError(f"drift must be a boolean, got {type(drift).__name__}")
+        return Verdict(drift=drift, reason=str(data["reason"]))
     except (json.JSONDecodeError, KeyError, TypeError) as e:
         raise CovenantError(f"judge returned an unparseable verdict: {raw!r}") from e

tests/test_fingerprint.py

@@ -2,6 +2,7 @@
 
 import pytest
 
+from covenant.errors import CovenantError
 from covenant.fingerprint import fingerprint, probe_key
 
 
@@ -30,7 +31,7 @@ def test_arrays_keep_items_only_when_uniform():
 
 
 def test_non_json_value_is_loud():
-    with pytest.raises(ValueError):
+    with pytest.raises(CovenantError):
         fingerprint({1, 2})