10 stories
How to configure OpenCode permission rules to keep child agents within their assigned git worktree — external_directory guard, bash restrictions, and per-agent overrides.
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.
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.
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.
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.
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
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
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
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.
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
Showing 10 of 10 stories
You're all caught up.
Get the top AI news and tools straight to your inbox every week.
Weekly · AI news from Grass Blog
Join 15,000+ AI enthusiasts
