PerformanceMonitor v2.5.0 released
See [CHANGELOG.md](https://github.com/erikdarlingdata/PerformanceMonitor/blob/main/CHANGELOG.md#250---2026-03-30) for details.
Releases, movers, new tools, and trending activity — sourced from GitHub signals.
See [CHANGELOG.md](https://github.com/erikdarlingdata/PerformanceMonitor/blob/main/CHANGELOG.md#250---2026-03-30) for details.
## What's New ### New Tools - **`browser_network_state_set`** — Toggle network offline mode for testing connectivity scenarios ([#39459](https://github.com/microsoft/playwright/pull/39459)) - **`browser_video_chapter`** — Add a chapter marker to the video recording, showing a f
## 0.8.20 — 2026-03-30 ### Security - Remove partial API key from health output (067) - Harden MCP allowlist for mutating study download paths (068) ### Fixes - `batch --json` always includes `_meta` contract field (069) - Harden GWAS variant section against live decode failu
FastMCP 3.2 is the Apps release. The 3.0 architecture gave you providers and transforms; 3.1 shipped Code Mode for tool discovery. 3.2 puts a face on it: your tools can now return interactive UIs — charts, dashboards, forms, maps — rendered right inside the conversation. ## Fast
## What's new **`get_repo_outline` now shows 2-level directory breakdown for large repos.** When a repository has more than 500 indexed files, the `directories` field now groups by two path components — `src/api/`, `src/models/`, etc. — instead of collapsing everything into a s
<!-- Release-Triggered-By: amirejaz --> ## What's Changed * Add scoped and user secret providers with system key isolation by @amirejaz in https://github.com/stacklok/toolhive/pull/4229 * Respect global config for OTel tracing and metrics enabled by @gkatz2 in https://github.com
## What's New ### Cross-Repository Dependency Tracking - **** — new tool: build a full dependency map across all indexed repos, showing which repos depend on which via package-level import matching - **** — new boolean parameter: extend importer search across all indexed repos
<!-- Release-Triggered-By: JAORMX --> ## What's Changed * Wire embedded auth server into VirtualMCPServer converter and deployment by @jhrozek in https://github.com/stacklok/toolhive/pull/4383 * Implement upstream_inject strategy and SubjectProviderName by @jhrozek in https://gi
## What's new ### Tri-state `use_ai_summaries` `use_ai_summaries` now accepts `"auto"` (new default — auto-detect from env keys), `true` (use explicit `summarizer_provider` + `summarizer_model`), or `false` (disable). Existing boolean configs work unchanged. ### `summarizer_mod
### Added - Session expiry controls for `StreamableHTTPTransport` via `session_idle_timeout:` option (#268) ### Changed - `ServerSession` for per-connection state (#275) ### Removed - Remove `Server#notify_progress` broadcast API (#276) - Remove undocumented handler override
## What's Changed * fix(logging): enhance error serialization in logs for better detail retention by @samanhappy in https://github.com/samanhappy/mcphub/pull/700 * fix(smart-routing): harden embedding under low RPM limits and surface failures to users by @josepcrespo in https://g
## What's Changed * Fix GitHub link in PR Review Inbox for desktop app by @Mng-dev-ai in https://github.com/Mng-dev-ai/agentrove/pull/438 * Add keyboard shortcuts for command menu by @Mng-dev-ai in https://github.com/Mng-dev-ai/agentrove/pull/439 * Add sidebar toggle, theme switc
## What's new ### Adaptive language config (languages_adaptive) New config key that auto-manages the languages list in .jcodemunch.jsonc based on languages actually detected during index_folder. Two modes: - Global config, no local config: on first index_folder, copies global c
## What's new **MiniMax and GLM-5 summarizer support** (PR #184 — contributed by SkaldeStefan) jcodemunch now supports the same multi-provider summarizer pattern as jdocmunch: - Set `MINIMAX_API_KEY` to auto-detect MiniMax M2.7 - Set `ZHIPUAI_API_KEY` to auto-detect GLM-5 - Us
## Bug fix **Gemini `CODE_RETRIEVAL_QUERY` KeyError on legacy `google-generativeai`** v1.12.4 introduced task-aware Gemini embeddings, but `CODE_RETRIEVAL_QUERY` was added in the newer `google-genai` SDK and is absent from the legacy `google-generativeai` package. Passing that
## What's new **YAML and Ansible parser support** — contributed by @SkaldeStefan (PR #183) `.yaml` and `.yml` files are now fully indexed with symbol extraction. ### Generic YAML Any `.yaml` / `.yml` file gets structural symbol extraction: - Container keys (dicts, lists) → `t
## Added * Added PowerShell execution logic extraction into a clean, static `PowerShellExecutor` class for improved code organization and stability (#158) * Added MSIX App path resolutions to support natively launching MSIX / UWP Windows applications for AI agents (#126) * Added
## What's new **Task-aware embedding for Gemini** (closes #181 — thanks @Aliuoly for the excellent suggestion!) When using Gemini embeddings, jcodemunch now passes the correct task type hint to the model: - `RETRIEVAL_DOCUMENT` when indexing symbols (via `embed_repo` or lazy e
## What's new ### Bug fixes **Cross-process LRU cache invalidation** — If you use the watcher (`jcodemunch-mcp watch`), you may have noticed agents falling back to grep/bash after editing files even though the watcher said it had re-indexed. Root cause: SQLite WAL mode does not
## What's new ### Added: `.razor` (Blazor component) file support `.razor` files are now fully indexed alongside `.cshtml` views. The Razor symbol extractor has been extended to handle Blazor-specific directives: - **`@page` routes** — emitted as `constant` symbols so componen