issues Search Results · language:Dune language:Python language:JavaScript language:JavaScript linked:pr language:Python
Filter by
4.4M results
Problem
entity_profiles uses TEXT PRIMARY KEY for the entity column. SQLite TEXT comparison is case-sensitive by default, so
Josh and josh create two separate entity profiles.
Impact
Medium — entity ...
bug
severity:medium
buildingjoshbetter
Problem
engine.py:1429 has a key name mismatch that causes temporal search results to be silently dropped or mishandled. The
temporal layer partially runs but its results may not integrate correctly into ...
bug
severity:medium
Problem
engine.py:409-450: The Qwen3 NaN fix runs inside _ensure_connection() on every first call for Base/Pro on macOS. It
drops ALL vector tables and re-embeds the entire database synchronously.
On ...
bug
severity:medium
Goal
Prevent duplicate Kafka message processing and ensure in-flight messages are fully processed before service shutdown for
production reliability.
Context
Affects the kafka-collector service (Golang, ...
engineering
status:merged
Lincyaw
Problem
HyDE (Hypothetical Document Embeddings) query expansion is supposed to be Pro-tier only. But the current code triggers
HyDE for any tier that has an API key configured, including Edge and Base. ...
bug
severity:high
Problem
Lock files in ~/.truememory/ persist indefinitely after the holding process dies:
- ingest.lock: 45 days old (since 2026-04-23)
- .spawn.lock: 25 days old (since 2026-05-12)
These may be ...
bug
severity:high
Problem
engine.delete(memory_id) removes the row from messages and messages_fts but does NOT clean up:
1. vec_messages (vector embedding)
2. vec_messages_sep (separation vector)
3. entity_profiles ...
bug
severity:high
Problem
Second-run search on the same database is 15x SLOWER than first-run search:
- Cold search (first run): 0.44s
- Warm search (second run): 6.84s
The warm run produces a different warning: ...
bug
severity:high
Problem
Model server allocates 2.33 GiB MPS GPU memory, exceeding the 1.95 GiB watermark limit. During batch embedding (200+
memories), MPS OOM triggers:
MPS backend out of memory (MPS allocated: 2.33 ...
bug
severity:critical
The problem
Since 2026.6.0 the Gardena Bluetooth integration is not reconnecting to the device. I get the error message unable to
find product type .
img width= 1040 height= 171 alt= Image src=
https://github.com/user-attachments/assets/9a08dcd7-1b8d-4718-aadb-b3c94b6dfec2 ...
integration: gardena_bluetooth
vonBrizien
Learn how you can use GitHub Issues to plan and track your work.
Save views for sprints, backlogs, teams, or releases. Rank, sort, and filter issues to suit the occasion. The possibilities are endless.Learn more about GitHub IssuesProTip! Restrict your search to the title by using the in:title qualifier.
Learn how you can use GitHub Issues to plan and track your work.
Save views for sprints, backlogs, teams, or releases. Rank, sort, and filter issues to suit the occasion. The possibilities are endless.Learn more about GitHub IssuesProTip! Restrict your search to the title by using the in:title qualifier.