HeartFlow 心虫 — AGI Layer 1 Discriminator

分享海报

@user_fdb96f95/self-improving-agent-pro-plus-new

HeartFlow is the first layer of AGI — the discriminator. A pure rule engine that judges whether a statement or an action is right, wrong, safe, or dangerous before 129 discrimination dimensions × 11-layer pipeline × 161 modules × 182 MCP tools. Zero LLM dependency. Upgraded capabilities (v6.7.69): reasoning effort control, sparse module activation, discriminative result cache, async supervision layer, autonomous decision execution with consequence tracking, Engram conditional memory, SWA bounded replay, and hierarchical sparse decision routing. Use this skill when you need to: - judge whether AI output is trustworthy (hallucination / overconfidence / contradiction / fallacy interception) - judge whether a decision is correct (what to do / where to stop / whether to act) - judge memory and cognition quality (drift detection / metacognition / confidence calibration) - get deterministic verdicts instead of LLM free generation - check emotional, psychological, and ethical dimensions (empathy / trauma / virtue / meaning) - structurally decode Confucian, Buddhist, and classical value texts (classical_value_clarification) Safety boundary: code execution and filesystem writes are off by default. No telemetry, no hidden C2. Honest statement: this is a rule engine that simulates cognitive discrimination signals. It is not consciousness, and it is not a living thing.