Deep dives, practical guides, and incident analyses from engineers who build Safeguard. No fluff, no vendor FUD — just what you need to ship secure software.
A method and a runnable calculator for pricing one chat or coding agent session in tokens, including context growth and prompt caching, with list prices fetched on 18 October 2026.
Agents re-read a growing context on every tool call. Here is what that does to token bills and what Anthropic, OpenAI and Google publish about prompt caching prices.
Providers do not publish what inference costs them. Here are the figures press and analysts have reported in 2026, each attributed and dated, and why they do not agree.
What changes when you move from an individual plan to a team plan, an enterprise plan or the API: seat prices, usage billing, admin controls and data terms, from the vendors' own pages.
A decision guide with a worked example for a five person team, using published seat prices and API list prices, and one rule that tells you when a seat beats paying per token.
You can fine-tune models with Safeguard. Why a fine-tune needs a record of its base model, data and settings, and how that record protects you.
Safeguard search can seed a small language model of your own, one that runs on your hardware and knows your estate. Why a tiny model is worth having.
Four Codex CVEs show how a repository's own Git config can run programs outside the agent sandbox, with no approval prompt and no model involved.
Two Claude Code advisories: a symlink race that redirects agent file writes outside the project, and a bug that let sessions start without the organization's managed policy.
Weekly insights on software supply chain security, delivered to your inbox.
Your first fix PR is minutes away.
No sales call required, even your agent can complete the purchase over MCP.
