hive-sh4re: drop the nix explanation from the panic message too

Same layering rule as the previous commit, one construct over. The
remediation text explained how nix wires the var and pointed at
`nix/assets.nix` -- a path that does not exist either, so it carried
both defects this branch is about.

What a developer hitting this panic needs is the layout to point at,
which is a fact about this crate.

hive-sh4re/src/assets.rs

@@ -3,9 +3,9 @@
 //! (`hive-c0re`) and the in-container harness binaries so they agree
 //! on the lookup contract.
 //!
-//! At runtime, the path is read from `$HIVE_ASSETS_DIR`. In nix
+//! At runtime, the path is read from `$HIVE_ASSETS_DIR`. Setting it is
-//! builds that env var is set by the `hive-c0re` / `harness-base` modules
+//! the deployment's job; this crate's contract is only that it names a
-//! to `${pkgs.hyperhive-assets}/share/hyperhive` (see `nix/assets.nix`).
+//! directory laid out like the packaged assets.
 //! For `cargo run` outside nix, set it yourself:
 //!
 //! ```sh
@@ -34,12 +34,10 @@ fn dir() -> PathBuf {
     match std::env::var("HIVE_ASSETS_DIR") {
         Ok(v) if !v.is_empty() => PathBuf::from(v),
         _ => panic!(
-            "HIVE_ASSETS_DIR is not set. Inside the nix-built systemd \
+            "HIVE_ASSETS_DIR is not set. A deployment sets it; when \
-             units this is wired automatically from \
+             running a binary outside one, point it at a directory \
-             `pkgs.hyperhive-assets`; for `cargo run` outside nix, \
+             with `branding/` and `prompts/` subdirs. See \
-             set it to a directory laid out like \
+             hive-sh4re/src/assets.rs for the contract.",
-             `nix/assets.nix`'s output (branding/ + prompts/ \
-             subdirs). See hive-sh4re/src/assets.rs for the contract.",
         ),
     }
 }

nix/agent-modules/weston-vnc.nix

@@ -6,7 +6,7 @@
 }:
 let
   # GUI processes run as the agent's own non-root user — the same user
-  # hive-agent runs as (declared + home-chowned by harness-base.nix) — so
+  # hive-agent runs as (declared + home-chowned by ./user.nix) — so
   # weston, the wayland client, and the agent share one user session.
   # `hyperhive.user.name` is set per-agent by the meta-flake renderer.
   userName = config.hyperhive.user.name;

nix/host-modules/hive-ci.nix

@@ -280,7 +280,7 @@ in
           # sandboxing always fails. Fall back to unsandboxed builds.
           # Moot once every nix invocation in the container routes
           # through the host daemon (the daemon governs sandboxing).
-          # See docs/gotchas.md and harness-base.nix.
+          # See docs/gotchas.md and nix/agent-modules/default.nix.
           nix.settings.sandbox-fallback = lib.mkForce true;
           # Degrade to a local build when a remote builder is unreachable
           # rather than failing the check. `fallback` is a client-side

nix/packages/default.nix

@@ -119,8 +119,8 @@ let
       '';
 
   # Per-bin split packages. Agent containers depend on the individual
-  # bins they actually exec/PATH-need (see harness-base.nix) instead
+  # bins they actually exec/PATH-need (see nix/agent-modules/packages.nix)
-  # of the `default` bundle — that keeps `hivectl` (dials the *host*
+  # instead of the `default` bundle — that keeps `hivectl` (dials the *host*
   # admin socket, unreachable from inside a container, drags in
   # `wireguard-tools`) and redundant binaries out of every agent's
   # closure.