Neuroboost Elixir
WarnAudited by ClawScan on May 10, 2026.
Overview
The included JavaScript is mostly local demo/algorithm code, but the skill’s instructions advertise persistent memory, background patrols, and automatic self-repair without clear boundaries or reviewed implementation.
Install only if you intentionally want persistent agent memory and self-management. Do not allow cron jobs, restarts, Telegram alerts, cleanup/deletion, or broad file/process scans until the exact scripts are present and reviewed, paths are restricted, backups are enabled, and human approval is required for high-impact actions.
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.
The agent could change or discard its own task or memory state, or try to restart itself, without you reviewing each action.
These are automatic state and process mutations, but the visible artifacts do not specify approval prompts, exact file/process scope, backups, or rollback behavior.
Self-Healing Rules include Auto-clean P2/P3 memories, Close stale P2 tasks, and Attempt agent restart.
Keep self-healing opt-in, require confirmation for deletion/restart/task-closing actions, restrict it to a declared workspace, and provide dry-run plus backups.
The agent may attempt ongoing self-monitoring or maintenance outside the immediate task if these instructions are followed.
The skill advertises automatic activation and recurring background health patrols, but the registry/install metadata does not clearly declare persistence or how the user controls these recurring actions.
auto-activate: true ... Quick Check (Heartbeat Mode) — Every 6-12 hours ... Bash and Node.js scripts for nightly cron jobs ... Auto-trigger on AHS < 60
Disable automatic patrols by default; require explicit setup, visible schedules, easy shutdown, and clear logs for any recurring job.
Sensitive or stale information could persist across sessions and influence future agent behavior, even after the original context is gone.
The skill promotes persistent storage of tasks, lessons, preferences, and identity across sessions, including a never-delete tier, without clear bounds on what may be stored, retention controls, or poisoning safeguards.
Task persistence — `.issues/` directory survives any context reset ... Three-layer memory — Daily logs → INDEX.md → MEMORY.md ... P0 (never delete), P1 (until superseded), P2 (30-day TTL).
Store only user-approved information, scope memory paths, define retention and deletion rules, review memory before reuse, and avoid treating stored notes as authoritative instructions.
A user or agent may rely on unseen automation for memory cleanup, health checks, or repairs, making the actual behavior unclear.
These high-impact helper scripts are referenced by the skill but are not present in the supplied file manifest, so their behavior cannot be reviewed here.
Supporting Scripts: `scripts/self-healing.js` — Main self-healing engine; `scripts/memory-distill.sh` — Memory distillation automation; Integrated into `health-quick-check.js` — Auto-trigger on AHS < 60
Include all referenced scripts in the package, align versions across SKILL.md/package/_meta/registry, and declare capabilities and persistence requirements explicitly.