
a9hec🔫😼PRMR
1.3K posts




Konu tartışmaya açık. Arda çok iyi oyuncu. Türk futbol tarihinin en büyük futbolcusu olma yolunda ilerliyor ama benim gözümde arda turan ve tugay kerimoğlu Arda Güler'den daha önde. Arda Turan en üst seviyede 6-7 sene oynadı. Neler neler yaptı Atletico'da. Messi neymar Suarez ile oynadı. Tugay 10 seneden fazla banko oynadı. Nihat kahveci fenomen Ronaldo ile gol krallığında yarıştı, real sociedad'ı real Madrid ile şampiyonluk yarışına soktu. Milli takımda ceklere neler yaptı. Arda güler'in önünde bu oyuncuları geçmesi için daha çok zamanı var. Geçecektir de. Geçmeli de. Ama büyük ünvanlar zaman gerektirir. Arda'nın daha zamana ihtiyacı var. Kıyaslamak ya da rol biçmek yerine destek olmak ve gurur duymak gerek




Many are wondering "what Google saw" that caused them to revise their post-quantum cryptography transition deadline to 2029 last week. It was this: research.google/blog/safeguard…

OpenFang v0.5.7 is out! here's everything since v0.4.9: - Argon2id password hashing for dashboard auth with random salts, replaces plain SHA-256, new openfang auth hash-password CLI command - SearXNG search provider, self-hosted metasearch with configurable instance URL, category validation, pagination, and noise filtering - SSRF allowlist, self-hosted/K8s users can configure ssrf_allowed_hosts in config.toml so agents can reach internal services while metadata endpoints stay blocked - image generation pipeline (DALL-E/GPT-Image) - WeCom channel adapter - Docker sandbox runtimes - shell skill runtime - release-fast build profile - wildcard tool capabilities, agent manifests can use mcp_filesystem_* patterns instead of enumerating every MCP tool - expanded embedding auto-detection, probes OpenAI, Groq, Mistral, Together, Fireworks, Cohere before local providers - XML tool call recovery format #14 for Llama-family models - Ollama context window defaults bumped to 128K context / 16K output - fix Qwen/OpenAI-compat tool_calls orphaning after context overflow, smart drain boundaries + streaming repair - fix LINE webhook signature validation, raw bytes for HMAC - fix local skill install TUI parsing, hot-reload via /api/skills/reload - fix exec_policy mode=full now bypasses approval gate for shell_exec - fix install script now correctly fetches latest release version - fix session endpoint returns full tool results (removed 2000-char truncation) - fix agent_send/agent_spawn timeout increased to 600s (was 120s) - fix model switching respects explicit provider via find_model_for_provider() - remove 3 decommissioned Groq models - fix Ollama streaming parser handles both reasoning_content and reasoning fields - fix model fallback chain retries with fallback_models on ModelNotFound instead of panicking - fix Gemini streaming SSE parser handles \r\n line endings, no more infinite empty retry loop - fix tool_use.input always normalized to JSON object, fixes Anthropic API errors - fix custom model names preserved, user-defined models take priority over builtins - fix heartbeat skips idle agents that never received a message, no more crash-recover loops - fix hand continuous interval changed from 60s to 3600s, prevents credit waste - fix global ~/.openfang/skills/ loaded for all agents, workspace skills properly override globals - fix openfang doctor reports all_ok=false when provider key is rejected - fix doctor --json outputs clean JSON to stdout - fix skill install detects Git URLs and clones before installing - fix workflows page scrollable - fix model dropdown handles object options, no more [object Object] for Ollama - fix spawn wizard provider dropdown loads dynamically from /api/providers (43 providers) - fix chat streaming renders in real-time (Alpine.js splice reactivity + stale WS guard) - fix [SILENT] token handling, case-insensitive matching, no longer stored in session history - fix token estimation includes ToolUse arguments in text_length() - fix Alpine.js settings page expression errors when budget data is null - fix agent skills hot-reload for TOML changes - fix Telegram startup timeout prevents daemon boot hang on flaky Local Bot API - invisible approval requests now visible with history, badge, and polling - Matrix auto_accept_invites now configurable, defaults to false for security - KaTeX loaded on demand to prevent first-paint blocking - Browser Hand checks python3 before python on modern Linux - 9 community PRs merged, 59 PRs merged total, 33 bugs fixed - full CSO security audit completed - 2,186+ tests passing, 0 clippy warnings github.com/RightNow-AI/op…











gm! We’re lining up something for CUBE holders. How many CUBEs do you have?








OpenFang: The Rust-Powered Agent OS Will Soon Be Taking Over The Internet search.app/rzHzG











