Self-Healing Agent
v1.0.0Self-recovery and auto-repair system for OpenClaw agents. Monitors agent health, detects failures (crashed cron jobs, broken skills, config corruption, memor...
⭐ 0· 47·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
medium confidencePurpose & Capability
The name/description (self-recovery, cron/memory/config/session checks) matches the included script's behavior: it reads/writes ~/.openclaw files, queries and restarts jobs via the 'openclaw' CLI, archives memory files, and attempts JSON fixes. No unrelated credentials, binaries, or install artifacts are requested.
Instruction Scope
SKILL.md tells the agent to run the shipped Python script (check/heal/monitor/report). The script performs file I/O under ~/.openclaw, can truncate/archive files, restart cron jobs via subprocess calls, and (per SKILL.md) may kill zombie sessions. Those actions are within the stated purpose but are high‑impact: they modify config and session files and invoke system commands. SKILL.md also promises escalation to the main session and network checks — verify how escalation is implemented in the remaining code before enabling autonomous heals.
Install Mechanism
No install spec; this is instruction-only with a bundled Python script. Nothing is downloaded or installed automatically by the registry metadata, so no third‑party install URLs or archives to audit.
Credentials
The skill requests no environment variables or credentials. It operates on local OpenClaw paths (~/.openclaw) and calls the local 'openclaw' CLI. This access is proportionate to a self‑healing agent. Note: if the code invokes git, remote network calls, or other CLIs (not fully visible in the truncated file), that could broaden its needs — review the remainder of the script for any network/credential usage.
Persistence & Privilege
always:false and default autonomous invocation are set (normal). The monitor mode runs continuously and will auto-apply fixes if invoked; that autonomous ability plus the script's capability to modify files and processes increases blast radius. There is no evidence it modifies other skills' configurations or requests permanent registry-level privileges.
Assessment
This skill appears coherent for auto-recovery of an OpenClaw agent, but it performs high-impact actions (editing configs, truncating/archiving files, restarting jobs, killing sessions). Before enabling automatic 'heal' or running as a monitor: 1) Review the entire scripts/self-healing-agent.py file (the provided snippet is truncated) for any network calls, remote endpoints, or commands that could exfiltrate data. 2) Run the script in --dry-run mode and inspect logs (use 'check' first). 3) Back up ~/.openclaw (config and memory) and test in a staging environment. 4) Prefer manual or moderated use of --aggressive fixes; avoid enabling autonomous continuous monitor without an audit. 5) Note the skill's source/homepage is unknown — lack of publisher info reduces trust; consider requesting provenance or a signed release from the author before granting this skill broad operational use.Like a lobster shell, security has layers — review code before you run it.
latestvk976enkgpmzj987kenzgf59xes8495w6
License
MIT-0
Free to use, modify, and redistribute. No attribution required.