CritterWatch 1.1, Bobcat 1.0 and AI Skills 1.17 ship; Stoat preview arrives
JasperFx shipped a coordinated Critter Stack release wave: CritterWatch 1.1, Bobcat 1.0, AI Skills 1.17 and a Stoat 0.5.0 preview. CritterWatch's MCP server now exposes 83 tools, and Bobcat 1.0 turns existing test methods into executable specifications. Worth a look if you run Wolverine or Marten in production or write integration tests on the .NET stack.
Key points
- CritterWatch 1.1 ships 83 MCP tools; the 13 irreversible ones answer their first call with a preview and a single-use confirmation token.
- AddCritterWatchMcp(o => o.ReadOnly = true) limits an agent to the 37 read-only tools.
- Wolverine 6.34 adds ScheduleRecurring for cron-based messages, and CritterWatch lists and pauses them.
- Each monitored service adds Wolverine.CritterWatch; the console itself installs exactly one store package: Marten/PostgreSQL, Polecat/SQL Server or Fisher/SQLite.
- Bobcat 1.0 is MIT licensed; add [Given], [When] and [Then] attributes to xUnit, TUnit, NUnit or MSTest tests.
- Stoat 0.5.0 is a public preview of a design-time surface for AI-assisted development across repositories.
JasperFx has shipped a coordinated Critter Stack release wave: CritterWatch 1.1, Bobcat 1.0, AI Skills 1.17 and a Stoat 0.5.0 preview. CritterWatch 1.1 is the monitoring console for Wolverine, Marten, Polecat and Fisher systems. Its MCP server now offers 83 tools, 37 of them read-only, and actions that cannot be undone require a preview and a single-use confirmation token. Fleet-wide Schedule Explorer and Document Explorer views are new, and much of the release is scalability work driven by very large production fleets.
Bobcat 1.0 is the first stable release of an MIT-licensed integration testing tool and the successor to Storyteller. Tests written with xUnit, TUnit, NUnit or MSTest serve as the specification: [Given], [When] and [Then] attributes mark steps on plain methods, and a Roslyn source generator records them at compile time. There is no separate DSL or binding layer, so readable specs evolve alongside the test suite you already maintain.
Real-world use
- An on-call team running a Wolverine fleet lets an agent triage dead letters while destructive replays wait for human confirmation.
- A multi-tenant Marten store watches per-tenant subscription lag and rewinds from one console.
- A service backed by a database and a broker gets readable integration specs that domain experts can review.
Why it matters
If you run Wolverine or Marten in production, the console now gives you per-tenant lag views, schedule control and agent-safe actions. If you write integration tests, Bobcat turns existing test methods into specifications without a second file.