Coding-agent telemetry now shows who acted and what was stopped
Copilot can now attribute activity to a user without capturing content, and Claude Code logs stopped tool calls. What to collect and how to read it.
Keeping personal and company data out of places it shouldn't go.
Copilot can now attribute activity to a user without capturing content, and Claude Code logs stopped tool calls. What to collect and how to read it.
Anthropic's API can now send OpenTelemetry spans to your own backend. What the content modes capture, how delivery works, and why token counts need care.
NCSC guidance and Copilot's sandbox proxy point the same way: agents should not hold real secrets. How credential proxies work, their limits, and a checklist.
5 min read
GitHub now blocks Anthropic API keys at push by default and is previewing AI-based secret detection. What changed, what is still preview, and what to do.
5 min read
GitHub Copilot CLI can discover models from a local Ollama instance. A local model is not offline mode, and telemetry and remote providers still need settings.
The UK ICO's call for evidence on agentic AI runs 8 October to 20 November 2026. Map security, transparency and ADM evidence now.
4 min read
AI coding tools can export prompts, tool arguments and full API bodies. Why it's off by default, the risks of switching it on, and how to decide.
Add two or three SpanDock guardrails for secrets, personal data and your own patterns, enforce them on every computer, and check that they work.