issues Search Results · language:Dune language:Python language:Java language:Java language:HTML language:CSS language:Java
Filter by
44M results
When users provide ambiguous input, LLMs typically either silently pick an interpretation or ask generic clarifying
questions — both failure modes that compound downstream errors. Recent work (Ask-before-Plan, ...
ambiguity
clarification
research
Spec Reference
Source: specs/fleet-usage-monitor.md Sections: §2.1, §3, §4.2, §4.3, §4.4, §4.5, §8 step 1
What to Build
Implement claude-config/scripts/usage_collector_claude.py — the Claude transcript ...
build-slice
fleet-usage-monitor
from-spec
P1
tests
The intent gap — the distance between an informal natural language requirement and a checkable formal specification — is
identified as the central challenge for reliable AI-generated code (arXiv:2603.17150), ...
formalization
high-impact
research
Existing benchmarks like SWE-bench measure whether an agent can fix a reported bug, but not whether the fix introduces
regressions — a core real-world concern given that maintenance consumes 70–90% of ...
regression
research
testing
Agentic coding workflows now average 1–3.5M tokens per task including retries and self-correction loops, and reasoning
models like DeepSeek-R1 consume substantially more tokens than non-reasoning counterparts ...
cost-efficiency
metrics
research
Before opening
- [x] I searched existing issues for a similar report.
- [x] This is not a security vulnerability.
Affected area
App UI
Mini EQ version
0.8.6
Install method
Flathub
Linux distribution ...
bug
Current pass@k metrics treat any test-passing solution as correct, ignoring that 45% of AI-generated code samples fail
OWASP Top 10 security tests and AI-generated code introduces 2.74× more vulnerabilities ...
high-impact
research
security
Reported by mahdation422 via Discord on 2026-06-06 (backfilled) Tags: Crash
Describe the Crash game crash when i go to war as usa
To Reproduce See Discord thread for details.
Screenshots No screenshots ...
crash
from-discord
politics
Problem
The dashboard reads each worker s :8080/1/summary and authenticates a direct miner with only two schemes
(build/dashboard/mining_dashboard/client/xmrig_client.py):
1. no token, or
2. Authorization: ...
dashboard
documentation
enhancement
security
setup

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.