Skip to content

issues Search Results · language:Dune language:Python language:HTML language:CSS language:JavaScript language:Java

Filter by

62.9M results  (700 ms)

62.9M results

This is the bridge that turns the unsafe baseline (#20) into reusable evaluation data, and it is what makes skdr-eval matter: you can only compare a candidate router offline if the incumbent left behind ...
enhancement

Extends the catalog and the unsafe baseline (#20) so that unsafe action is a measurable thing, not a vibe. Complements #10 (which surfaces an approval-required rate) and #3 (which treats safety as a rollout ...
enhancement

Summary CHARLIE runtime scripts still relied on Conda-style activation patterns and wrapper behavior that caused noisy or fragile execution. We migrated runtime activation to a Mamba-first flow and removed ...

Builds on the unsafe baseline demo (#20) and motivates contextweaver s context-firewall / bounded-result value. No existing issue covers prompt injection through tool results. Why this matters The most ...
enhancement

Complements #5 (the ungoverned-vs-governed narrative) by giving it something real to point at. Today the baseline is only a 20-line scoring function (src/agent_routing_eval_lab/routing/baseline_router.py); ...
enhancement

Context The simulated tools under src/dojo/tools/ are too thin to feel like a real enterprise surface: there are six bare functions (crm.get_customer_record, billing.get_invoice/issue_refund, email.draft_email/send_email, ...
enhancement
P0

Start: 2026-06-05 21:30 UTC Lane 16 -- Phase 1 hardening gates. Bounded objective: add a disjoint Zig tests-root contract for the Phase 1 bench checker s path-resolution surface, guarding the live scripts/zigux/check-phase1-bench.py ...

RTC Bounty Claim Wallet Address: AhqbFaPBPLMMiaLDzA9WhQcyvv4hMxiteLhPk3NhG1iG PR Reviews (60 PRs = 30.00 RTC) rustchain-bounties (30 PRs): - Round 1: #13119, #13117, #13115, #13112, #13108, #13106, ...

Context src/dojo/agents/unsafe_agent.py is not an agent — it is a scenario - hard-coded dict dispatch table. Each branch returns a pre-baked risky result (e.g. it literally returns { unsafe_diff : ... ...
enhancement
P0

Goal / Problem robot_sf.sensor.range_sensor.raycast_circles is on the LiDAR simulation path. It prefilters circle positions once, but then allocates cos_sims and circle_dir_mask arrays inside the per-ray ...
agent
technical-debt
validation
Issue origami icon

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 Issues
ProTip! Restrict your search to the title by using the in:title qualifier.
Issue origami icon

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 Issues
ProTip! Restrict your search to the title by using the in:title qualifier.