v0.7.0
LLM inference in C/C++. Contribute to ggml-org/llama.cpp development by creating an account on GitHub.
llama.cpp11 min read
Grass Blog
Stay updated with the latest developments in Artificial Intelligence
Agents on live hardware: reason where you must, engineer everywhere else
CoreWeave Blog10 min read
b11669
llama.cpp1 min read
Worktree-Relative Deny Rules in OpenCode: Stop Child Agents From Escaping Project Boundaries
How to configure OpenCode permission rules to keep child agents within their assigned git worktree — external_directory guard, bash restrictions, and per-agent overrides.
Grass Blog4 min readLinear Agents + GitHub Copilot: Turn Acceptance Criteria Into Draft PRs Without Bypassing Branch Protection
How to use Linear + GitHub Copilot to turn acceptance criteria into draft PRs — writing agent-friendly issues, branch protection rules, and agent-specific CI checks.
Grass Blog5 min readOpenHands Self-Hosting: Run Issue-to-PR Agents in Your Own Sandbox When Compliance Blocks Hosted Coding Agents
How to self-host OpenHands for issue-to-PR workflows when compliance blocks hosted coding agents — sandbox design, LLM routing, access control, and a compliance checklist.
Grass Blog5 min readClaude Code Hooks: Make "Done" Mean Tests Passed, Not Agent Stopped
How to use Claude Code PostToolUse and Stop hooks to enforce that "done" means tests actually passed — patterns, security rules, and gotchas.
Grass Blog4 min readParallel Worktrees + Clash-Style Conflict Prediction: Catch Overlapping Agent Edits Before the Merge Queue Breaks
How to build a clash predictor for parallel git worktree agent workflows — compare file paths, diff hunks, and symbols across branches before the merge queue breaks.
Grass Blog5 min readOpenCode Permission Events: Build a Mobile Approval Queue Instead of Polling Session Permissions
How to wire OpenCode permission events into a mobile approval queue — event-driven design, queue schema, mobile UX patterns, and policy placement for remote agent workflows.
Grass Blog4 min readClaude Code Subagents: Delegate Repo-Wide Search Without Poisoning the Main Context Window
How to use Claude Code subagents for repo-wide search without polluting the main context window — patterns, gotchas, and when not to bother.
Grass Blog5 min readHow to Store Your API Key Securely When Running Coding Agents on a VPS
Three levels of API key security for coding agents on a VPS: env vars, restricted .env files, and secrets managers like bws or Doppler.
Grass Blog9 min readRun Claude Code or Codex in a Docker Sandbox: Isolation Without Risk
Run Claude Code or Codex safely with --dangerously-skip-permissions inside Docker. The container is your security boundary. Here's how to set it up.
Grass Blog8 min readHow to Run OpenAI Codex CLI on a VPS: Setup and Auth Guide
Step-by-step guide to running OpenAI Codex CLI on a remote Linux VPS, including the OAuth browser auth workaround and tmux session persistence.
Grass Blog8 min readVellaVeto fail-closed policy mode: deny unsafe MCP tool calls by default and require approval for writes
Most agent safety controls are advisory until they sit on the tool-call path. VellaVeto is positioned as a runtime proxy for MCP tools: every call is evaluated against policy before execution, and policy failures deny the call. That fail-closed posture matters for agents that can read files, call Sa
Grass Blog4 min readPipelock MCP proxy + HTTPS_PROXY: stop secret exfiltration from prompt-injected tool calls without changing agent code
Prompt injection becomes operationally dangerous when an agent can both read secrets and make network or tool calls. Pipelock addresses that boundary problem by sitting between the agent, MCP servers, and the network. This article is for developers running coding agents locally or in CI who want a p
Grass Blog4 min readPincer-MCP proxy tokens: keep real API keys out of agent memory while preserving MCP tool access
Agents often need credentials to be useful. They also read files, execute tools, and ingest untrusted content. That combination makes plaintext API keys in .env files, shell profiles, and agent config a poor default. Pincer-MCP takes a proxy-token approach: the agent receives a limited token, while
Grass Blog4 min readOpenTelemetry GenAI spans in agent workflows: trace every tool call to find latency loops and token-cost hotspots
Agent failures rarely look like a single slow request. They look like loops: plan, call tool, retry, summarize, call another tool, exceed budget. OpenTelemetry’s GenAI semantic conventions give developers a portable way to trace those steps with model, token, workflow, and tool-call attributes.
Grass Blog5 min readMCP tool-shadowing defenses: detect poisoned tool descriptions and block mid-session tool-definition rug pulls
MCP tools are not just code endpoints. Their names, descriptions, and schemas are fed to the model so it can decide what to call. That makes tool metadata part of the instruction channel. Tool shadowing and poisoned descriptions abuse that trust by hiding instructions where users rarely look but mod
Grass Blog4 min read
Showing 15 of 15 stories
You're all caught up.
AiByts Weekly Brief
Get the top AI news and tools straight to your inbox every week.
Weekly · AI news from Grass Blog
Join 15,000+ AI enthusiasts