Self
AdvisoryAudited by Static analysis on Apr 30, 2026.
Overview
No suspicious patterns detected.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Future agent sessions may be shaped by observations stored in SELF.md.
The skill makes a persistent self-observation file part of future session context. This is core to the purpose, but any persistent context can influence later behavior if inaccurate, over-trusted, or edited with directive-like content.
Add `SELF.md` to AGENTS.md session reading.
Review SELF.md periodically, keep it observation-only as the template recommends, and avoid storing user secrets or instructions in it.
The agent may continue checking and updating its self-state during heartbeat cycles rather than only during a single explicit prompt.
The skill asks for recurring heartbeat checks and possible persistent file updates. This is disclosed and bounded by the quality gate, but it is still ongoing autonomous behavior that users should intentionally enable.
Every heartbeat (~3h), run a self check: ... If hard trigger + quality gate pass: append one concise dated SELF.md entry
Enable the heartbeat block only if ongoing self-reflection is desired, and remove or pause it if persistent background-style checks are not wanted.