You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: CHANGELOG.md
+19Lines changed: 19 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -4,6 +4,25 @@ All notable changes to Mnemolis are documented here, from v3.45.0 onward. For ev
4
4
5
5
---
6
6
7
+
## [3.56.1]
8
+
9
+
### Fixed — fusion no longer carries `history`'s honest abstentions as noise
10
+
11
+
Found while CLI-verifying v3.56.0 for release: *"what are the latest AI trends"* and *"tell me about the history of ancient rome"* correctly avoided the keyword triggers (those regression pins held), but the **LLM fusion picker included the `history` source anyway** — its own NAME attracts it for those phrasings — and its plain-prose *"I couldn't identify which sensor or metric you mean…"* abstention was fused into otherwise-good answers as a trailing noise block.
12
+
13
+
Fixed with the project's established two-part pattern (the discourse-framing precedent explicitly rejects description-nudging as the sole lever):
14
+
15
+
- The source **description is corrected where it was genuinely inaccurate** — it now states it covers THIS house's own sensors and services only, NOT world history, general knowledge, news, or public/industry trends.
16
+
- The **verifiable lever**: `history`'s five honest-abstention messages ("couldn't identify which sensor", "couldn't identify which kind of event", the ambiguity ask, "requires temporal pattern detection", "no recorded metrics yet") joined `_looks_empty()`'s canonical phrase list, so fusion drops them exactly as it drops any other member with nothing to contribute. The `**` structural guard means no real history answer (headers are always `**{name} — …**`) can ever match; the plain-prose event-count answers ("1 opening today, …") were verified to survive the new phrases, with pins.
17
+
18
+
Field-verified before tagging: both genre queries re-run against a live deployment came back with no trailing `[HISTORY]` block, on both fresh phrasings and the original cached-then-expired ones.
19
+
20
+
### Changed
21
+
-**Dependency: `mcp[cli]` 1.27.2 → 1.28.1** (Dependabot). Full test suite — including the MCP transport tests — run against 1.28.1 before packaging, not just trusted from the PR.
22
+
- Version bumped to 3.56.1. Test suite: **1697 passing** (3 new pins for the abstention/`_looks_empty` contract), ruff clean. One stale comment fixed (`cache_ttl_history_seconds` referenced the removed sampling-interval setting).
23
+
24
+
---
25
+
7
26
## [3.56.0]
8
27
9
28
### Added — the `history` source: time-series memory for the house (opt-in, off by default)
description="Unified local knowledge search API with multi-source fusion. Routes queries to Kiwix, Open-Meteo, FreshRSS, SearXNG, Uptime Kuma, or multiple sources concurrently.",
"uptime": "Uptime Kuma monitor status. Use when asked about service status, what is down, or network health.",
428
428
"ha": "Home Assistant entity states. Use for house status summaries, which lights are on, door and lock status, battery levels, indoor sensors, motion events, or power consumption.",
429
429
"changes": "Snapshot diff engine. Use when asked what changed, any new outages, weather changes, or new news since a given time.",
430
-
"history": "Recorded history of the house's own sensors and services — temperature, CO2, humidity, power, service uptime, and door/motion events over time. Use for questions about past values, highs/lows, averages, counts, or trends.",
430
+
"history": "Recorded readings from THIS house's own sensors and services only — temperature, CO2, humidity, power, service uptime, and door/motion events over time. Use for questions about this home's past sensor values, highs/lows, averages, counts, or sensor trends. NOT for world history, general knowledge, news, or public/industry trends.",
431
431
"fusion": "Multi-source fusion — queries multiple sources concurrently and merges results. Use for complex queries that benefit from combining offline knowledge, live web, and recent news.",
0 commit comments