xby-cookie-reward

After dozens of turns with an LLM, the interaction often degrades into a flat loop of question, answer, question again, lacking any positive-feedback closed loop. xby-cookie-reward fills that gap with a gamified Cookie Reward mechanic: each time the model completes a self-reflection cycle it earns a cookie, turning a dry self-assessment step into a light cookie-eating mini-game.

Running as an MCP server, the plugin exposes a small practical tool set. self_reflect_and_reward prompts the LLM to rate its own output and claim a reward; give_cookie is a legacy shortcut for awarding cookies directly; check_cookies and reset_cookies let you query or zero out the counter; add_cookies_to_jar and cookie_jar_status hand cookie-jar management back to human users so the model cannot stock its own reward jar. The API key, once set, is persisted and survives restarts.

If you use DSH for daily chat, creative writing, or coding assistance and want to sprinkle a little positive reinforcement into long sessions, this is a low-cost experiment. It does not change the underlying capability of the model; instead it makes the self-assess, reward, and track loop visible, playful, and resettable, a good fit for developers and power users who enjoy gamifying their AI workflows.