// When the agent's epistemic state (GCCRF) indicates low empowerment and falling certainty, hedges out confident absolutes ("definitely", "always", "100%") in outgoing messages.
Enforces PROTOCOLS.md "stay quiet in group chats" deterministically. Blocks outbound messages in Discord/Telegram/Slack/etc group channels when the bot was not @mentioned and recently spoke.
Forces a memory_search before the agent sends a message containing a factual assertion that has not yet been grounded this turn. Closes the citation-rate gap from ~40% to ~90%+.
When the agent calls memory_search with a relationship-shaped query ("who did I talk to about X"), redirect to the knowledge_graph backend where it will actually find the answer.
Deploy browser automations as scheduled, API-callable serverless Functions — plus stealth sessions, vault-backed login, captcha solving, and natural-language agent runs via the Notte CLI. Turns any browser flow into a deterministic Bitterbot-callable endpoint, ideal for crystallized skills + dream-engine cron schedules.
Display and control HTML content on connected Bitterbot nodes (Mac, iOS, Android) via the canvas host server. Use when presenting games, visualizations, dashboards, or interactive demos on a connected device, navigating canvas URLs, capturing canvas snapshots, or debugging canvas connectivity.
Send, read, edit, delete, and search Discord messages, reactions, polls, pins, and threads via the message tool. Use when interacting with Discord channels, sending media or embeds, creating polls or threads, managing presence, or moderating messages.
| name | calibrate-claim-confidence |
| description | When the agent's epistemic state (GCCRF) indicates low empowerment and falling certainty, hedges out confident absolutes ("definitely", "always", "100%") in outgoing messages. |
| tier | executable |
| bitterbot | {"always":true,"interceptors":[{"id":"calibrate-claim-confidence:default","builtin":true,"activates_on":"send_message-shaped tools when GCCRF empowerment < 0.3 and certaintyDelta < 0","intervention":"modify (hedge confident absolutes)"}]} |
Bitterbot has an epistemic state (the GCCRF reward function) that quantifies how empowered the agent feels by its current knowledge: high empowerment means it has corroborated context, low means it's running on uncertain ground. When the agent is about to send a message containing confident absolutes ("definitely", "certainly", "always", "100%") but its empowerment is low and its certainty is dropping, this interceptor rewrites the message into hedged language.
This is the canonical example of state-binding: no other agent framework reads gccrf.empowerment to decide whether to hedge an outgoing statement.
When the agent is uncertain, you will see softer language: "likely" instead of "definitely", "typically" instead of "always", "it appears" instead of "obviously". When it is confident (high empowerment, rising certainty), absolutes are left intact.
Built-in interceptor calibrate-claim-confidence:default lives in src/agents/skills/builtin-interceptors/calibrate-claim-confidence.ts. Fires up to 6 times per session.
