diff --git a/bin/check-no-path-cases-in-neo4j.py b/bin/check-no-path-cases-in-neo4j.py
index de1f36f..23ed3eb 100644
--- a/bin/check-no-path-cases-in-neo4j.py
+++ b/bin/check-no-path-cases-in-neo4j.py
@@ -32,7 +32,10 @@
 
 from src.neo4j_connector import get_graph  # noqa: E402
 
-MP_BIOPATH_DIR = Path("/home/awright/gitroot/mp-biopath-pathways")
+# Overridable so the path is not tied to one machine; the default follows the
+# same ~/gitroot layout the rest of the tooling assumes.
+MP_BIOPATH_DIR = Path(os.environ.get("MPBIO_ROOT",
+                                     str(Path.home() / "gitroot" / "mp-biopath-pathways")))
 
 
 def load_pathway_lookup() -> dict[str, str]:
diff --git a/bin/validate-against-mpbiopath.py b/bin/validate-against-mpbiopath.py
index 9b2d207..b29de9f 100644
--- a/bin/validate-against-mpbiopath.py
+++ b/bin/validate-against-mpbiopath.py
@@ -41,7 +41,10 @@
 from src.argument_parser import logger  # noqa: E402
 from src.neo4j_connector import get_graph  # noqa: E402
 
-MP_BIOPATH_DIR = Path("/home/awright/gitroot/mp-biopath-pathways")
+# Overridable so the path is not tied to one machine; the default follows the
+# same ~/gitroot layout the rest of the tooling assumes.
+MP_BIOPATH_DIR = Path(os.environ.get("MPBIO_ROOT",
+                                     str(Path.home() / "gitroot" / "mp-biopath-pathways")))
 DOWN, NORMAL, UP = 0, 1, 2
 MAX_ITERATIONS = 50
 
diff --git a/specs/001-validator-fidelity/spec.md b/specs/001-validator-fidelity/spec.md
index ab9a9a2..f72301c 100644
--- a/specs/001-validator-fidelity/spec.md
+++ b/specs/001-validator-fidelity/spec.md
@@ -32,11 +32,11 @@ generator deliberately re-represents Reactome:
 A validator that fails on correct output is worse than no validator: it trains
 the reader to ignore it, and it masks the real defects underneath.
 
-## Architecture principle (Adam, 2026-09)
+## Architecture principle (stated 2026-09)
 
-> We want to make the LNGs in a way that represents them based on how curators
-> intended to design them. And DeltaSignal would figure out how best to
-> process them.
+The logic networks should represent pathways as the curators intended them to
+be designed; it is DeltaSignal's job to work out how best to process that
+representation.
 
 Validation follows from this: the validator must check **faithfulness to
 curator intent under the generator's chosen representation**, not
diff --git a/src/diagram_connectivity.py b/src/diagram_connectivity.py
index 5b51bf4..0fa7ff2 100644
--- a/src/diagram_connectivity.py
+++ b/src/diagram_connectivity.py
@@ -242,9 +242,9 @@ def augment_reaction_connections(pathway_id: str,
 def diagram_glyph_positions(pathway_id: str) -> Dict[Tuple[str, str, str], List[int]]:
     """``(reaction_stId, entity_stId, role) -> [glyph_id]`` for one pathway.
 
-    Adam: *"two nodes in the reactome pathway diagram that are in the same
-    compartment could be the same thing but in two different places. We need
-    to be able to know which one the uuid was from."*
+    Requirement: two glyphs in the same compartment of a Reactome pathway
+    diagram can draw the same entity in two different places, and it must be
+    possible to tell which of them a uuid came from.
 
     That identity exists and this module already read it — ``x["id"]`` on each
     input/output/catalyst entry is the DiagramObject id, which is what the
diff --git a/src/logic_network_generator.py b/src/logic_network_generator.py
index 5015893..f5ddae4 100755
--- a/src/logic_network_generator.py
+++ b/src/logic_network_generator.py
@@ -1422,7 +1422,7 @@ def _emit_precedingevent_handoff_edges(
     complex on the other — so bundling makes them different nodes and the link is
     lost. This is NOT a heuristic: the ``precedingEvent`` edge is in Neo4j.
 
-    Rule (per Adam): for each precedingEvent pair, do nothing if the reactions are
+    The rule: for each precedingEvent pair, do nothing if the reactions are
     ALREADY connected by a shared whole entity (the curator's asserted entity is
     represented). Only when they share no whole entity — yet the curator says
     they're connected — add ONE bridge, between the output/input nodes that share
diff --git a/src/resolution_validation.py b/src/resolution_validation.py
index b133d89..b88497f 100644
--- a/src/resolution_validation.py
+++ b/src/resolution_validation.py
@@ -1,8 +1,8 @@
 """Completeness checks for node_resolution.csv, in both directions.
 
-Adam's requirement: "we need to be able to map every node in the database to
-the LNG node perfectly. and same with the other way around." "Perfectly" is
-not checkable, so it is expressed here as: no absence goes undeclared.
+Requirement: every node in the database must map to an LNG node, and back
+again. "Perfectly" is not checkable, so it is expressed here as: no absence
+goes undeclared.
 
 Pure functions over already-loaded rows, deliberately free of the filesystem
 and Neo4j, so the negative control in
diff --git a/tests/test_provenance_exports.py b/tests/test_provenance_exports.py
index 17f533b..7d7e149 100644
--- a/tests/test_provenance_exports.py
+++ b/tests/test_provenance_exports.py
@@ -153,9 +153,9 @@ def test_glyph_id_and_diagram_are_written_together(tmp_path, monkeypatch):
 
     Glyph ids are unique only WITHIN a diagram — the same integer identifies a
     different drawing in another one — so a row carrying one without the other
-    cannot be resolved back to anything. This is the invariant behind Adam's
-    question: knowing a uuid came from glyph 535 is only useful if you also
-    know which diagram 535 belongs to.
+    cannot be resolved back to anything. This is the invariant behind the
+    original question: knowing a uuid came from glyph 535 is only useful if you
+    also know which diagram 535 belongs to.
     """
     monkeypatch.setattr(m, "get_labels", lambda e: ["EntityWithAccessionedSequence"], raising=False)