Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

nowledge-mem-openclaw-plugin/package.json (1)

39-42: Validate openclaw.release flags in the validator too.

You now declare openclaw.release.publishToClawHub / publishToNpm on Line 39-Line 42, but they are not currently enforced by scripts/validate-plugin.mjs, so accidental drift won’t fail validation.

♻️ Suggested validator patch

diff --git a/nowledge-mem-openclaw-plugin/scripts/validate-plugin.mjs b/nowledge-mem-openclaw-plugin/scripts/validate-plugin.mjs
@@
   assertString(openclaw.install?.npmSpec, "package.json openclaw.install.npmSpec");
   assertVersionFloor(openclaw.install?.minHostVersion, "package.json openclaw.install.minHostVersion");
   assertVersionFloor(openclaw.compat?.pluginApi, "package.json openclaw.compat.pluginApi");
   assertString(openclaw.build?.openclawVersion, "package.json openclaw.build.openclawVersion");
+  if (typeof openclaw.release?.publishToClawHub !== "boolean") {
+    fail("package.json openclaw.release.publishToClawHub must be a boolean");
+  }
+  if (typeof openclaw.release?.publishToNpm !== "boolean") {
+    fail("package.json openclaw.release.publishToNpm must be a boolean");
+  }

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@nowledge-mem-openclaw-plugin/package.json` around lines 39 - 42, Add
validation in scripts/validate-plugin.mjs to enforce the new openclaw.release
flags: check that package.json (or parsed plugin manifest) contains
openclaw.release.publishToClawHub and openclaw.release.publishToNpm and that
they are booleans; if missing or not boolean, throw or log an error and exit
non‑zero so validation fails. Locate the existing validation flow (e.g. the
validatePluginConfig / validateManifest function in scripts/validate-plugin.mjs)
and insert a small check that reads manifest.openclaw?.release and asserts
typeof publishToClawHub === "boolean" and typeof publishToNpm === "boolean",
emitting a clear error message referencing openclaw.release.publishToClawHub /
openclaw.release.publishToNpm when validation fails.

nowledge-mem-openclaw-plugin/scripts/validate-plugin.mjs (1)

111-123: Broaden configSchema validation beyond three defaults.

Current checks only protect sessionContext, sessionDigest, and corpusSupplement. Key/type/default/range drift in the remaining schema properties won’t be caught by this validator.

Suggested direction

   const configSchema = manifest.configSchema?.properties;
   if (!configSchema || typeof configSchema !== "object") {
     fail("openclaw.plugin.json configSchema.properties must exist");
   }
+  const requiredConfigKeys = [
+    "sessionContext",
+    "sessionDigest",
+    "digestMinInterval",
+    "maxContextResults",
+    "recallMinScore",
+    "maxThreadMessageChars",
+    "captureExclude",
+    "captureSkipMarker",
+    "corpusSupplement",
+    "corpusMaxResults",
+    "corpusMinScore",
+    "apiUrl",
+    "apiKey",
+  ];
+  for (const key of requiredConfigKeys) {
+    if (!(key in configSchema)) {
+      fail(`configSchema.properties.${key} must exist`);
+    }
+  }
   if (configSchema.sessionContext?.default !== false) {
     fail("sessionContext must default to false");
   }

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@nowledge-mem-openclaw-plugin/scripts/validate-plugin.mjs` around lines 111 -
123, The validator currently only checks three defaults; update the logic around
manifest.configSchema?.properties (configSchema) to validate the full expected
schema set: ensure required property keys exist, their schema types match
expectations, and their default values/ranges match a defined spec instead of
only checking sessionContext, sessionDigest, and corpusSupplement; implement a
small expectedSpec mapping (property -> {type, default, optional range}) and
iterate over Object.entries(expectedSpec) to call fail(...) when a property is
missing, has the wrong type, or the default/value is outside the allowed range,
reusing the existing fail function for errors.

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

nowledge-mem-cursor-plugin/README.md (1)

49-57: Good separation of concerns, minor clarity suggestion.

This section correctly distinguishes between:

• Cursor MCP settings (for plugin tool calls)
• Local nmem client config (for terminal-side bootstrap and handoffs)

The nmem config client set commands are accurate and align with the learnings about remote Mem setup.

✨ Minor clarity improvement for line 56

Consider rephrasing line 56 for better clarity:

-Cursor MCP settings cover the plugin's tool calls. The local `nmem` client config covers the terminal-side bootstrap and handoff behaviors. It is separate from server-side remote-access settings on the Mem host.
+Cursor MCP settings cover the plugin's tool calls. The local `nmem` client config covers the terminal-side bootstrap and handoff behaviors. This client-side config is separate from the server-side configuration that governs remote access on the Mem host itself.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@nowledge-mem-cursor-plugin/README.md` around lines 49 - 57, Reword the
sentence that starts "Cursor MCP settings cover the plugin's tool calls..." to
make the separation clearer: state that Cursor MCP settings control only plugin
tool calls (server/plugin-side), while the local `nmem` client config controls
terminal-side bootstrap and handoff behaviors (client-side), and explicitly note
that server-side remote-access settings live on the Mem host and are configured
separately; reference the terms "Cursor MCP settings", "local `nmem` client
config", "sessionStart hook", and "save-handoff skill" so the reader knows which
behaviors each configuration affects.

nowledge-mem-hermes/client.py (1)

177-199: Consider chaining FileNotFoundError with from None.

The static analysis correctly flags that the FileNotFoundError exception should use raise ... from None to avoid misleading exception chains. The JSONDecodeError handling at line 196-199 already does this correctly with from error.

♻️ Proposed fix

         except FileNotFoundError:
-            raise RuntimeError(
+            raise RuntimeError(
                 "nmem CLI not found. Install: pip install nmem-cli, "
                 "or enable CLI in Nowledge Mem: Settings > Developer Tools"
-            )
+            ) from None

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@nowledge-mem-hermes/client.py` around lines 177 - 199, The FileNotFoundError
except block handling the subprocess.run call should suppress the original
exception chain; update the except FileNotFoundError branch that raises
RuntimeError("nmem CLI not found...") to use exception chaining suppression by
raising the RuntimeError from None (i.e., use "raise RuntimeError(...) from
None") so the misleading FileNotFoundError is not shown; locate the block around
the subprocess.run call that references self._timeout and modify that raise
accordingly.

nowledge-mem-hermes/provider.py (1)

380-404: _parse_csv and _normalize_id_list are nearly identical—consider consolidating.

Both methods have the same implementation: handle None, split strings by comma, handle iterables, and convert other types to string. Consider extracting a shared helper or having one call the other.

♻️ Proposed consolidation

     `@staticmethod`
-    def _parse_csv(value: Any) -> Optional[List[str]]:
+    def _normalize_list(value: Any) -> Optional[List[str]]:
+        """Normalize value to a list of trimmed strings, or None if empty."""
         if value is None:
             return None
         if isinstance(value, str):
             return [item.strip() for item in value.split(",") if item.strip()]
         if isinstance(value, (list, tuple, set)):
             parsed = [str(item).strip() for item in value if str(item).strip()]
             return parsed or None
-
         text = str(value).strip()
         return [text] if text else None

+    # Alias for semantic clarity in different contexts
+    _parse_csv = _normalize_list
+    _normalize_id_list = _normalize_list
+
-    `@staticmethod`
-    def _normalize_id_list(value: Any) -> Optional[List[str]]:
-        if value is None:
-            return None
-        if isinstance(value, str):
-            return [item.strip() for item in value.split(",") if item.strip()]
-        if isinstance(value, (list, tuple, set)):
-            parsed = [str(item).strip() for item in value if str(item).strip()]
-            return parsed or None
-
-        text = str(value).strip()
-        return [text] if text else None

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@nowledge-mem-hermes/provider.py` around lines 380 - 404, The two almost
identical methods _parse_csv and _normalize_id_list should be consolidated:
extract a single helper (e.g., _to_str_list or _parse_list_like) that
encapsulates the logic of handling None, splitting strings on commas,
normalizing iterables (list/tuple/set) by str().strip(), and converting other
values to a single-item list or None, then have both _parse_csv and
_normalize_id_list call that helper (or remove one and alias it) to avoid
duplication and ensure consistent behavior across parse/normalize operations.

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".