Merge pull request #1340 from oraios/dev-dj

Improvements for next patch release

Author: opcode81

CHANGELOG.md

@@ -5,9 +5,14 @@ Status of the `main` branch. Changes prior to the next official version change w
 * General:
   - Support environment variable `SERENA_USAGE_REPORTING` (set to `false` to disable usage reporting)
   - Extended the list of always ignored directories (by language servers) with common cases.
+  - Improve exposed toolset: With mode switching no longer being a feature, we now fully apply tool exclusions 
+    defined by modes when in a single-project context (limiting exposed tools to a minimum)
   - Fix: When scanning for `.gitignore` files, the presence of files that could not be made relative 
     to the project root would cause the scan to fail. #1317
 
+Dashboard:
+  - Fix handling of read news, saving each read news entry separately #1338
+
 JetBrains: 
   - Improve handling of `relative_path` parameter 
      - Improve its documentation to avoid usage errors
@@ -18,7 +23,7 @@ JetBrains:
   - Add mSL (mIRC Scripting Language) support (custom pygls-based language server; symbols, references, definitions)
   - Fix initialisation issues in Vue language server #1333
 
-# 1.1.1
+# v1.1.1 (2026-04-12)
 
 * General:
   - Enable cert verification for HTTPS request to oraios-software.de #1320
@@ -30,7 +35,7 @@ JetBrains:
   - Fix Dart LSP returning only symbol name as body instead of full method body.
 
 
-# 1.1.0
+# v1.1.0 (2026-04-11)
 
 * General:
   - **Major**: Add commands for hooks and documentation of recommended setup. Consider setting up the [recommended hooks](https://oraios.github.io/serena/02-usage/030_clients.html) !
@@ -53,7 +58,7 @@ JetBrains:
     Some clients would terminate the MCP server in a way that did not ensure proper termination.
   - Fix: Manual server shutdown triggered by GUI tool/dashboard not cleaning everything up.
 
-# 1.0.0
+# v1.0.0 (2026-04-03)
 
 * General:
     * Add monorepo/multi-language support
@@ -134,7 +139,7 @@ JetBrains:
   * **C/C++ alternate LS (ccls)**: Add experimental, opt-in support for ccls as an alternative backend to clangd. Enable via `cpp_ccls` in project configuration. Requires `ccls` installed and ideally a `compile_commands.json` at repo root.
   * **Add support for Solidity** via the Nomic Foundation `@nomicfoundation/solidity-language-server` (automatically installed via npm)
 
-# 0.1.4
+# v0.1.4 (2025-08-15)
 
 ## Summary
 
@@ -178,7 +183,7 @@ Fixes:
   default shell reconfiguration imposed by Claude Code)
 * Additional wait for initialization in C# language server before requesting references, allowing cross-file references to be found.
 
-# 0.1.3
+# v0.1.3 (2025-07-22)
 
 ## Summary
 

news/20260414.html

@@ -0,0 +1,15 @@
+<div class="news-item">
+    <h3>Agents Evaluating Serena for Themselves & Maintenance Update</h3>
+    <p class="date">April 14, 2026</p>
+    <p>
+        <b>Evaluation.</b>
+        We had our actual "end users", i.e. coding agents, evaluate the added value of Serena's tools based on their own empirical insights.
+        Read what the agents had to say in our <a target="_blank" href="https://oraios.github.io/serena/04-evaluation/000_evaluation-intro.html">evaluation report</a>.<br>
+        This represents the first systematic evaluation we ever conducted.
+    </p>
+    <p>
+        <b>Maintenance Update.</b> The new v1.1.2 release adds several improvements and fixes.<br>
+        If you have already switched to the uv tool installation of Serena, upgrade as follows:<br>
+        uv tool upgrade serena-agent --prerelease=allow
+    </p>
+</div>

scripts/bump_version.py

@@ -3,6 +3,7 @@
 import logging
 import os
 import re
+from datetime import datetime
 from pathlib import Path
 from typing import Literal
 
@@ -185,7 +186,9 @@ def update_changelog(changelog_text: str, new_version: str) -> str:
     unreleased_body = unreleased_section[len(_UNRELEASED_HEADER) :]
     intro, unreleased_entries = split_unreleased_body(unreleased_body)
 
-    updated_section = _UNRELEASED_HEADER + intro + f"# {new_version}\n"
+    date_str = datetime.now().strftime("%Y-%m-%d")
+    updated_section = _UNRELEASED_HEADER + intro + f"# v{new_version} ({date_str})\n"
+
     if unreleased_entries.strip():
         updated_section += "\n" + unreleased_entries.lstrip("\n")
     else:

src/serena/agent.py

@@ -453,31 +453,35 @@ def _create_base_toolset(
         tool_inclusion_definitions.append(serena_config)
         tool_inclusion_definitions.append(context)
 
+        # determine whether we are operating in a single-project context
+        # (i.e. the project that is activated at startup is the only project that will be worked with throughout the session)
+        is_single_project = context.single_project and project is not None
+
         # consider modes
-        # Since modes can be dynamically turned on and off, we don't include their definitions directly,
-        # For the initially active dynamic modes, we make sure that the tools they enable are included.
-        for mode in modes.get_default_modes():
-            tool_inclusion_definitions.append(
-                NamedToolInclusionDefinition(
-                    name=f"InitialDynamicModeInclusions[{mode.name}]", included_optional_tools=mode.included_optional_tools
-                )
-            )
-        # For the base modes, we also apply the tool exclusions, since they apply throughout the entire session
+        # * base modes: These cannot be changed, so they are fully applied
         for base_mode in modes.get_base_modes():
-            tool_inclusion_definitions.append(
-                NamedToolInclusionDefinition(
-                    name=f"BaseMode[{base_mode.name}]",
-                    included_optional_tools=base_mode.included_optional_tools,
-                    excluded_tools=base_mode.excluded_tools,
+            tool_inclusion_definitions.append(base_mode)
+        # * default modes: When not in a single-project context, these modes are dynamic (can later be turned off),
+        #   so we consider only their inclusions (but not their exclusions, because these must not be hard)
+        for mode in modes.get_default_modes():
+            if is_single_project:
+                tool_inclusion_definitions.append(mode)
+            else:
+                # Since modes can be dynamically turned on and off, we don't include their definitions directly,
+                # For the initially active dynamic modes, we make sure that the tools they enable are included.
+                tool_inclusion_definitions.append(
+                    NamedToolInclusionDefinition(
+                        name=f"InitialDynamicModeInclusions[{mode.name}]", included_optional_tools=mode.included_optional_tools
+                    )
                 )
-            )
 
         # When in a single-project context, the agent is assumed to work on a single project, and we thus
         # want to apply that project's tool exclusions/inclusions from the get-go, limiting the set
         # of tools that will be exposed to the client.
         # Furthermore, we disable tools that are only relevant for project activation.
         # So if the project exists, we apply all the aforementioned exclusions.
-        if context.single_project and project is not None:
+        if is_single_project:
+            assert project is not None
             log.info(
                 "Applying tool inclusion/exclusion definitions for single-project context based on project '%s'",
                 project.project_name,

src/serena/config/context_mode.py

@@ -53,7 +53,13 @@ def print_overview(self) -> None:
         """Print an overview of the mode."""
         print(f"{self.name}:\n {self.description}")
         if self.excluded_tools:
-            print(" excluded tools:\n  " + ", ".join(sorted(self.excluded_tools)))
+            print(" excluded tools:\n    " + ", ".join(sorted(self.excluded_tools)))
+        if self.included_optional_tools:
+            print(" included optional tools:\n    " + ", ".join(sorted(self.included_optional_tools)))
+        if self.fixed_tools:
+            print(" fixed tools:\n    " + ", ".join(sorted(self.fixed_tools)))
+        if self.prompt:
+            print(" defines initial prompt")
 
     @classmethod
     def from_yaml(cls, yaml_path: str | Path) -> Self:

src/serena/config/serena_config.py

@@ -83,7 +83,11 @@ def __init__(self) -> None:
         If a name of a mode matches a name of a mode in SERENAS_OWN_MODES_YAML_DIR,
         the user mode will override the default mode definition.
         """
-        self.news_snippet_id_file: str = os.path.join(self.serena_user_home_dir, "last_read_news_snippet_id.txt")
+        self.news_legacy_last_read_id_file: str = os.path.join(self.serena_user_home_dir, "last_read_news_snippet_id.txt")
+        """
+        file containing the ID of the last read news snippet
+        """
+        self.news_read_items_file: str = os.path.join(self.serena_user_home_dir, "news_read.pkl")
         """
         file containing the ID of the last read news snippet
         """

src/serena/dashboard.py

@@ -15,6 +15,7 @@
 from PIL import Image
 from pydantic import BaseModel
 from sensai.util import logging
+from sensai.util.pickle import dump_pickle, load_pickle
 
 from serena.analytics import ToolUsageStats
 from serena.config.serena_config import SerenaConfig, SerenaPaths
@@ -132,6 +133,54 @@ def from_task_info(cls, task_info: TaskExecutor.TaskInfo) -> Self:
         )
 
 
+class ReadNews:
+    def __init__(self, read_ids: list[str], legacy_last_read_id: str | None = None):
+        self._read_ids = set(read_ids)
+        self._legacy_last_read_id = legacy_last_read_id
+
+    @staticmethod
+    def load() -> "ReadNews":
+        read_news_path = SerenaPaths().news_read_items_file
+        legacy_last_read_id_path = SerenaPaths().news_legacy_last_read_id_file
+
+        def load_legacy_last_read_id() -> str | None:
+            if not os.path.exists(legacy_last_read_id_path):
+                return None
+            with open(legacy_last_read_id_path, encoding="utf-8") as f:
+                last_read_news_id = f.read().strip()
+                if last_read_news_id == "20262103":
+                    last_read_news_id = "20260321"  # fix originally misnamed news id
+                return last_read_news_id
+
+        if os.path.exists(read_news_path):
+            return load_pickle(read_news_path)
+        else:
+            instance = ReadNews(read_ids=[], legacy_last_read_id=load_legacy_last_read_id())
+            instance._save()
+            try:
+                os.unlink(legacy_last_read_id_path)
+            except:
+                pass
+            return instance
+
+    def _save(self) -> None:
+        dump_pickle(self, SerenaPaths().news_read_items_file)
+
+    def is_read(self, identifier: str) -> bool:
+        if identifier in self._read_ids:
+            return True
+        if self._legacy_last_read_id is not None and identifier <= self._legacy_last_read_id:
+            return True
+        return False
+
+    def mark_read(self, identifier: str) -> None:
+        """
+        Marks the given news snippet as read, saving the new state to disk
+        """
+        self._read_ids.add(identifier)
+        self._save()
+
+
 class SerenaDashboardAPI:
     log = logging.getLogger(__qualname__)
 
@@ -150,6 +199,7 @@ def __init__(
         self._loaded_news: dict[str, str] = {}
         self._news_ready = threading.Event()
         self._setup_routes()
+        self._read_news = ReadNews.load()
         # Fetch remote news in background on startup (non-blocking)
         threading.Thread(target=self._fetch_news, daemon=True).start()
 
@@ -362,25 +412,19 @@ def _fetch_unread_news() -> dict[str, str]:
                 self._news_ready.wait()
                 all_news = self._loaded_news
 
-                # Filter news items by installation date
                 serena_config_creation_date = SerenaConfig.get_config_file_creation_date()
                 if serena_config_creation_date is None:
                     # should not normally happen, since config file should exist when the dashboard is started
                     # We assume a fresh installation in this case
                     log.error("Serena config file not found when starting the dashboard")
                     return {}
                 serena_config_creation_date = serena_config_creation_date.strftime("%Y%m%d")
-                # Only include news items published on or after the installation date
+
+                # filter for news after the installation date
                 post_installation_news = {k: v for k, v in all_news.items() if k >= serena_config_creation_date}
 
-                news_snippet_id_file = SerenaPaths().news_snippet_id_file
-                if not os.path.exists(news_snippet_id_file):
-                    return post_installation_news
-                with open(news_snippet_id_file, encoding="utf-8") as f:
-                    last_read_news_id = f.read().strip()
-                    if last_read_news_id == "20262103":
-                        last_read_news_id = "20260321"  # fix originally misnamed news id
-                return {k: v for k, v in post_installation_news.items() if k > last_read_news_id}
+                # read unread news
+                return {k: v for k, v in post_installation_news.items() if not self._read_news.is_read(k)}
 
             try:
                 unread_news = _fetch_unread_news()
@@ -393,9 +437,7 @@ def mark_news_snippet_as_read() -> dict[str, str]:
             try:
                 request_data = request.get_json()
                 news_snippet_id = str(request_data.get("news_snippet_id"))
-                news_snippet_id_file = SerenaPaths().news_snippet_id_file
-                with open(news_snippet_id_file, "w", encoding="utf-8") as f:
-                    f.write(news_snippet_id)
+                self._read_news.mark_read(news_snippet_id)
                 return {"status": "success", "message": f"Marked news snippet {news_snippet_id} as read"}
             except Exception as e:
                 return {"status": "error", "message": str(e)}