Inner Life Core
ReviewAudited by ClawScan on May 10, 2026.
Overview
The skill is transparent about creating an agent memory and behavior loop, but its rules can let the agent act automatically based on persistent trust and habit state without clearly bounded action limits.
Install only if you intentionally want persistent agent memory and behavior shaping. Before using it with shell, browser, git, account, or other high-impact tools, edit BRAIN.md so automatic actions are narrowly scoped and require explicit approval for risky changes.
Findings (4)
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.
If the agent has powerful tools, it may make changes first and only report afterward once its stored trust or habit values cross a threshold.
These instructions authorize automatic behavior based on mutable local state, but do not define safe action boundaries, reversibility, or which tools/actions still require explicit user approval.
**Habits:** `habits.myHabits` with `strength >= 3` → follow automatically. **Trust routing:** - `technical >= 0.7` → fix yourself, report afterwards - `proactive >= 0.7` → act, then tell
Treat the BRAIN protocol as requiring explicit approval for file, account, shell, browser, spending, publishing, or configuration changes unless the user has separately approved a narrow action class.
Stored local notes and relationship or habit values can shape future agent decisions, including how proactive the agent becomes.
The skill intentionally creates and reuses persistent memory about emotions, habits, and relationship state across sessions.
reads: ["memory/inner-state.json", "memory/drive.json", "memory/habits.json", "memory/relationship.json", "BRAIN.md"] writes: ["memory/inner-state.json", "memory/drive.json", "memory/habits.json", "memory/relationship.json", "memory/daily-notes/"]
Periodically inspect and prune the memory files, especially relationship trust levels, habits, daily notes, and any files used by other inner-life skills.
Hidden-looking comment tags in notes could influence later agent behavior if they are inserted or modified unexpectedly.
The protocol uses HTML-comment signal tags in daily notes, which may be less visible in rendered Markdown but can still steer future loops.
Inter-component communication via daily notes (HTML comments).
Review raw Markdown for daily notes, not only rendered previews, and avoid treating signal tags as authoritative unless they were created by a trusted workflow.
The skill can become a persistent source of behavior instructions for long-running agent setups.
The skill does not install a scheduler, but it is designed to guide recurring or autonomous agent loops if the user connects it to such workflows.
> Central protocol for autonomous agent operation. > All scheduled loops reference this file.
Use it only in workspaces where you want persistent agent behavior, and keep scheduled or autonomous loops under explicit user-controlled limits.