Supervised Agentic Loop

Self-improving AI agent loop with built-in misalignment detection. An AI agent autonomously runs Brainstorm → Plan → Implement → Review → Evolve cycles — kee...

This package appears coherent for its stated purpose, but it runs code and modifies your repository: 1) Run it only in an isolated working directory or disposable git clone (it creates branches and will reset/rollback). 2) Review/limit the metric_command you pass — it will be executed (it can do anything the shell allows). 3) The agent you provide can be a local subprocess (MONITOR_LLM_COMMAND or an AgentCallable); that subprocess may itself perform network I/O — treat it as untrusted unless you control it. 4) Optional Telegram alerts require you to provide a bot token/chat id; leave those unset to prevent outbound alerts. 5) Inspect git_isolation and verification gate code before use to confirm rollback semantics and which files are checked. If you want tighter isolation, run inside a container or VM, or add stricter verification/read-only rules before enabling autonomous runs.

Type: OpenClaw Skill Name: supervised-agentic-loop Version: 0.1.2 The 'supervised-agentic-loop' bundle is a sophisticated framework for autonomous code optimization with a strong emphasis on security and alignment monitoring. It implements a multi-phase loop (Brainstorm, Plan, Implement, Review, Verify, Evolve) that uses git-based isolation for safe rollbacks and reputation scoring to suspend underperforming or 'hallucinating' agents. Most notably, the 'sal/monitor' subpackage provides a two-phase defense system: a synchronous prefilter that blocks dangerous commands (e.g., privilege escalation, persistence, and exfiltration) and an asynchronous LLM-based session review. It also includes a credential sanitizer (sal/monitor/sanitizer.py) to redact secrets from logs and a self-monitoring heartbeat with canary tests. All high-risk behaviors, such as command execution and optional Telegram alerting, are well-documented and strictly aligned with the tool's defensive and experimental purpose.