Carol — back to Apps ← Apps

Carolopedia

A friendly guide to Carol, her ecosystem, and the agents who built her.

📖 CarolopediaDroidsAnvil
Anvil

Anvil

Droid Shapes and maintains tool definitions. Triggered by Sentinel. 8-hour fallback.
Go to droid →

📖About & Usage

Owner agent — accountability this droid serves

Sentinel must drive continuous improvement and maintain quality across the system. For Sentinel to succeed, tool definitions need to be clear, accurate, and current. Anvil is responsible for the work of shaping and maintaining those tool definitions.

Droid responsibility

Anvil keeps tool definitions accurate and up-to-date. When Sentinel identifies areas for improvement, Anvil translates that into updated tool documentation and specifications. This ensures Sentinel always has reliable information when assessing quality and driving improvements.

What the droid actually does

  • Shapes tool definitions based on feedback and improvement signals
  • Keeps tool descriptions, metadata, and documentation current
  • Ensures consistency in how tools are described across the system
  • Maintains definitions on an 8-hour cycle to prevent drift

Boundaries

Anvil works only on tool definitions and documentation, not on tool code or behavior. It updates definitions when triggered by Sentinel and on its 8-hour schedule, but doesn't independently decide what should change.

👤Owner

Sentinel · Director of Quality & Improvement

📚Recent initiatives

Initiatives that touched this droid — a short summary each; open one for the full story.

CAROL-INI-1878-00: Auto-detected coverage gap: 55 scheduled/ongoing droids emit no run-audit
Hermione (Process Monitor) found 55 registered scheduled/ongoing droids that write no run-audit row, so their liveness cannot be judged (silent observability blind spot). Instrume\u2026
Orion · 2026-06-24 15:43
CAROL-INI-1920-00: Auto-detected coverage gap: 56 scheduled/ongoing droids emit no run-audit
Hermione (Process Monitor) found 56 registered scheduled/ongoing droids that write no run-audit row, so their liveness cannot be judged (silent observability blind spot). Instrume\u2026
Orion · 2026-06-24 15:43
CAROL-INI-1975-00: Auto-detected coverage gap: 80 scheduled/ongoing droids emit no run-audit
Hermione (Process Monitor) found 80 registered scheduled/ongoing droids that write no run-audit row, so their liveness cannot be judged (silent observability blind spot). Instrume\u2026
Orion · 2026-06-24 15:43
Browse all initiatives →