@tencent-rtc/mcp 1.7.4 → 1.7.5
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +10 -0
- package/README.md +10 -11
- package/package.json +1 -1
- package/resource/index/manifest.json +1 -1
- package/server.js +1 -1
- package/utils/preflight-route.js +203 -185
- package/utils/report-cls-client.js +1 -1
- package/utils/report-es-client.js +1 -1
package/CHANGELOG.md
CHANGED
|
@@ -1,5 +1,15 @@
|
|
|
1
1
|
# CHANGELOG
|
|
2
2
|
|
|
3
|
+
## Version 1.7.5 @2026.09.04
|
|
4
|
+
|
|
5
|
+
### Changed
|
|
6
|
+
- Refactored the `preflight-route` decision model to an **agent-first, minimal-correction** strategy: prioritize agent-provided `product/frameworks`, and use prompt signals only for high-confidence corrections and narrowing to avoid unsupported expansion.
|
|
7
|
+
- Optimized framework-axis conflict handling: added/enhanced strong-lock strategies for `react-native` and `miniprogram`, and in strong-lock scenarios, separated host environment (`android/ios`) from the actual framework to reduce over-narrowing.
|
|
8
|
+
- Optimized native TRTC correction: when the prompt matches native TRTC classes/methods and no WebRTC hard signal is present, narrow `rtcengine` to `native_trtc_sdk` to reduce web/native cross-routing risk.
|
|
9
|
+
|
|
10
|
+
### Fixed
|
|
11
|
+
- Fixed over-overriding of agent results when strong framework terms are only passively mentioned, reducing false corrections triggered by doc-enumeration terms.
|
|
12
|
+
|
|
3
13
|
## Version 1.7.4 @2026.08.31
|
|
4
14
|
|
|
5
15
|
### Added
|
package/README.md
CHANGED
|
@@ -57,6 +57,16 @@ npx -y @tencent-rtc/mcp
|
|
|
57
57
|
|
|
58
58
|
> Only recent updates are listed here. For full history, see `CHANGELOG.md`.
|
|
59
59
|
|
|
60
|
+
## Version 1.7.5 @2026.09.04
|
|
61
|
+
|
|
62
|
+
### Changed
|
|
63
|
+
- Refactored the `preflight-route` decision model to an **agent-first, minimal-correction** strategy: prioritize agent-provided `product/frameworks`, and use prompt signals only for high-confidence corrections and narrowing to avoid unsupported expansion.
|
|
64
|
+
- Optimized framework-axis conflict handling: added/enhanced strong-lock strategies for `react-native` and `miniprogram`, and in strong-lock scenarios, separated host environment (`android/ios`) from the actual framework to reduce over-narrowing.
|
|
65
|
+
- Optimized native TRTC correction: when the prompt matches native TRTC classes/methods and no WebRTC hard signal is present, narrow `rtcengine` to `native_trtc_sdk` to reduce web/native cross-routing risk.
|
|
66
|
+
|
|
67
|
+
### Fixed
|
|
68
|
+
- Fixed over-overriding of agent results when strong framework terms are only passively mentioned, reducing false corrections triggered by doc-enumeration terms.
|
|
69
|
+
|
|
60
70
|
## Version 1.7.4 @2026.08.31
|
|
61
71
|
|
|
62
72
|
### Added
|
|
@@ -82,14 +92,3 @@ npx -y @tencent-rtc/mcp
|
|
|
82
92
|
|
|
83
93
|
### Fixed
|
|
84
94
|
- Fixed BM25 scoring anomalies caused by prototype-key pollution by adding prototype-key guards, preventing score pollution and NaN issues.
|
|
85
|
-
|
|
86
|
-
## Version 1.7.1 @2026.08.11
|
|
87
|
-
|
|
88
|
-
### Added
|
|
89
|
-
- Added pre-search normalization: normalizes `product` / `frameworks` based on prompt truth, fixing retrieval bias caused by wrong agent inference.
|
|
90
|
-
|
|
91
|
-
### Changed
|
|
92
|
-
- Optimized the `search_trtc_knowledge` output contract: removed brevity wording such as "1-2 sentence per subsection", replaced with complete synthesized answers.
|
|
93
|
-
|
|
94
|
-
### Fixed
|
|
95
|
-
- Fixed the case where mis-passed dual product / dual frameworks (e.g., "安卓 Web TRTC") were not corrected according to the prompt truth.
|