Spaces:
Sleeping
Sleeping
Commit History
-
Enhance session log UI with compact copy option for Hugging Face Space iframes
Co-authored-by: OpenAI <noreply@openai.com>5ca5cbb -
Update README.md
Verified signaturec3dee41 -
Link published social post and video walkthrough in README
Co-authored-by: OpenAI <noreply@openai.com>9a8af0b -
Fix duplicate logs and add session download capability
abc5b16
-
Update README.md
Verified signature1b81b54 -
stabilize live button hover colors
OpenAI Codex +1: OpenAI CodexLock the Live GPU button palette across Gradio normal, hover, active, focus, and disabled states to prevent secondary-style color glitches. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>d1239dd -
Update README.md
VxkYUnverified signature3413671 -
clean public submission section
OpenAI Codex +1: OpenAI CodexReplace the rough README checklist with a concise submission assets table and ignore a local private submission tracker. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>5df4f52 -
trim README award caveats
OpenAI Codex +1: OpenAI CodexRemove the negative not-claimed award list from the hackathon README while preserving positive sponsor alignment. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>a2711b8 -
add README hero banner
OpenAI Codex +1: OpenAI CodexAdd a polished SVG hero banner and centered project links/badges so the repository landing page presents Vyber like a product page after the Hugging Face metadata block. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>d27a7d2 -
fix README architecture SVG layout
OpenAI Codex +1: OpenAI CodexClean up the static architecture diagram by removing overlapping labels, widening the verdict box, adjusting arrows, and preserving aspect ratio. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>897796f -
render README architecture diagram
OpenAI Codex +1: OpenAI CodexReplace the Mermaid architecture block with a static SVG so the Hugging Face file viewer displays the diagram reliably. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>182c708 -
hide paused backend internals
OpenAI Codex +1: OpenAI CodexKeep Modal/local fallback details in server logs while showing visitors a clean paused-GPU message. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>de60d1a -
clarify paused GPU state
OpenAI Codex +1: OpenAI CodexExplain that the live Modal GPU backend may be paused to save compute while demo replay remains available. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>c3d4a9b -
add MIT license
OpenAI Codex +1: OpenAI CodexAdd a standard MIT license file for the Vyber project. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>1ddafca -
upgrade hackathon README
OpenAI Codex +1: OpenAI CodexRewrite the Space card around Vyber's product story, current safety boundary, sponsor alignment, submission checklist, and future roadmap. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>3d48345 -
fix live launch button state
OpenAI Codex +1: OpenAI CodexKeep the Live GPU launch button readable during Gradio disabled/loading states by overriding its background, border, opacity, and text color. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>840b5fe -
restore Vyber landing design
OpenAI Codex +1: OpenAI CodexRestore the previous hero, metadata chips, note, status banner, and wide terminal layout while keeping demo/live actions as compact buttons. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>bdde775 -
simplify Vyber landing controls
OpenAI Codex +1: OpenAI CodexReduce hero copy, remove metadata chips and framed notes, and compact the scenario/action row for a quieter first screen. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>c4ede4e -
add public demo replay mode
OpenAI Codex +1: OpenAI CodexAdd a no-GPU demo replay path, a separate live GPU launch button, queue limits, and a live-run admission lock to protect Modal budget during public traffic. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>bf94d3c -
harden Red exploit report normalization
OpenAI Codex +1: OpenAI CodexEnsure every expected vulnerability appears in the exploit report, strip leaked action text from report fields, and share attack recipes with re-attack verification. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>4b8287b -
align harness wording with Vyber branding
OpenAI Codex +1: OpenAI CodexRemove user-facing OpenCode wording from backend comments and diagnostics while preserving required internal installer and config names. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>6a55c61 -
tighten sandbox trace accuracy
OpenAI Codex +1: OpenAI CodexClear stale sandbox files per run, keep Red recon tools deterministic inside the sandbox, strip ANSI terminal escapes, and clarify final Red verdict wording. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>ebc06b1 -
restore HTML operation terminal
OpenAI Codex +1: OpenAI CodexReturn the operation output to the full HTML terminal shell while keeping the native Gradio launch event binding. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>5c1211d -
restore launch event with native terminal autoscroll
OpenAI Codex +1: OpenAI CodexRemove custom page-level tracking JavaScript and switch the operation terminal to a Gradio textbox with built-in autoscroll so the launch button remains on the native event path. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>1ff50a5 -
allow Vyber harness sandbox tools
OpenAI Codex +1: OpenAI CodexRun OpenCode inside the sandbox with scoped permissions so the Blue agent can read, edit, grep, and execute commands during controlled cyber-range repairs. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>e8010d5 -
track live Vyber terminal output
OpenAI Codex +1: OpenAI CodexScroll the operation terminal into view on launch and keep it pinned to the latest streamed agent output. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>8d65ea2 -
refine Vyber hero presentation
OpenAI Codex +1: OpenAI CodexClean up the top hero copy, spacing, and runtime note for a calmer Gradio landing surface. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>7e2f4ca -
stabilize Vyber harness deployment
OpenAI Codex +1: OpenAI CodexUse the CUDA 12.1 llama-cpp-python wheel to avoid source builds, resolve the Vyber/OpenCode binary robustly inside Modal, and lock the launch button contrast. Built with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>1f4d914 -
fix launch button contrast
OpenAI Codex +1: OpenAI CodexBuilt with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>7ea2766 -
route blue remediation through Vyber harness
OpenAI Codex +1: OpenAI CodexBuilt with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>0a8f2ba -
verify patches through red re-attack
OpenAI Codex +1: OpenAI CodexBuilt with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>5026d80 -
increase terminal text contrast
OpenAI Codex +1: OpenAI CodexBuilt with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>5f24916 -
avoid repeated red-blue retry loops
OpenAI Codex +1: OpenAI CodexBuilt with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>ac82a04 -
stabilize operation terminal and agent validation
OpenAI Codex +1: OpenAI CodexBuilt with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>1ac998c -
calm operation terminal styling
OpenAI Codex +1: OpenAI CodexBuilt with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>185944c -
pin Gradio local runtime dependencies
OpenAI Codex +1: OpenAI CodexBuilt with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>ea52d9e -
polish Vyber cyber-range experience
OpenAI Codex +1: OpenAI CodexBuilt with OpenAI Codex for the Hugging Face Build Small Hackathon. Co-authored-by: OpenAI Codex <codex@openai.com>6b56e7b -
feat: loop-until-zero vuln engine — 3 vulns/scenario, multi-round Red+Blue agents, per-vuln scoreboard
7c7a303 -
feat: terminal-aesthetic output — timestamps, box headers, THINK/EXEC/STDOUT blocks, tight AI prompts
5305cf0 -
style: red+blue terminal themed colors - crimson tint vs navy tint, accent border strips
de88958 -
fix: comprehensive CSS glitch elimination - text shadows, dark bleed, readonly contrast
3de1755 -
revert: restore correct 7B GGUF filename
8f15d02 -
fix: update 7B GGUF filename to Q4_K_M suffix
6f6f1d6 -
Parameterize model size options and add 7B/1.5B tiered fallback search in backend
a77aed6 -
Rename tokenizer to processing_class in SFTTrainer for new TRL compatibility
026193b -
Rename max_seq_length to max_length for new TRL SFTConfig compatibility
d95d733 -
Enforce trl>=0.12.0 in training image
28f6689 -
Use SFTConfig for modern TRL SFTTrainer
6d38272 -
Fix dataset column names in training script
015ffbe