Pi 1.5.0 adds custom self-update commands, extension changelogs and GPT-6.1 Terra
The terminal coding agent released 1.5.0 on November 2. It also moves tool argument validation to its own schema package, a breaking change for extensions that used TypeBox-only features.
Pi released version 1.5.0 on November 2. pi update can now run your own update command (set PI_SELF_UPDATE_COMMAND or selfUpdateCommand, for example mise use -g npm:{packageName}@{version}). Installed extension packages can show their changelog on the next launch after an update. And large tool calls stream faster, because each delta is parsed once instead of re-parsing everything received so far.
GPT-6.1 Terra is available through OpenAI, Azure OpenAI Responses and OpenAI Codex, with a 500K context window on Codex. The default OpenAI model moves from gpt-5.5 to gpt-6.1-sol.
The breaking change is tool argument validation, which now runs on @earendil-works/pi-schema. TypeBox features outside JSON Schema, such as Type.Refine() and Type.BigInt(), no longer apply to tool arguments. Plain JSON Schema tools, MCP tools included, now get the coercion TypeBox tools already had, like turning numeric strings into numbers.
Three security fixes in the list: a saved 'Do not trust' decision is now respected for a project that only has an AGENTS.md, MCP sign-in rejects file: and other non-web OAuth URLs, and terminal titles can no longer inject escape sequences.
Why it matters for builders
If you wrote a Pi extension with TypeBox refinements on tool arguments, those checks stop running after you update. The trust fix closes a case where a folder with only AGENTS.md was auto-trusted against your saved choice.
Try this
Before you run pi update, search your extension code for Type.Refine, Type.BigInt, Type.Undefined and Format.Set. Any tool argument that relies on them needs its check moved into the tool body once you're on 1.5.0.
Sources
Get stories like this every Monday
One tested AI workflow with the prompts, plus 3 news items that matter if you build. Free.
More AI news
- Anthropic renames Max 20x to Ultra and drops the five-hour limit for verified accounts
- SemiAnalysis launches GDPx, an agent benchmark where the best model scores 12%
- DeepSeek open-sources dsh-superagent, a research prototype of a coding agent that takes work from GitHub mentions
- ChatGPT and Codex will share one token-based allowance on Plus, Pro and Business from December