Self Improving Agent
v1.0.2Local skill for capturing learnings, errors, corrections, and patterns to enable continuous agent improvement. Processes events locally in your OpenClaw agen...
⭐ 0· 529·2 current·3 all-time
byclaw0x@kennyzir
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
The name/description (self-improvement via local event processing) matches what is implemented: a TypeScript handler that accepts events, infers severity, extracts tags, generates insights and suggested rules. There are no unexpected required binaries, env vars, or external integrations declared.
Instruction Scope
SKILL.md instructs the agent how to submit events and how to use returned entries; the handler.ts implements only event validation, local analysis, and summary building. There are no instructions to read unrelated files, access other system state, or transmit data to external endpoints.
Install Mechanism
No install spec is present (instruction-only install via openclaw CLI), and the included handler.ts is pure local code. Nothing is downloaded from external URLs or installed to nonstandard locations.
Credentials
The skill declares no environment variables, no primary credential, and the code does not read process.env or any credential/config paths. Requested access is minimal and proportional to the stated functionality.
Persistence & Privilege
always is false; the skill is stateless by design and returns entries to the caller. It does not persist data, modify other skills, or require elevated/system-wide privileges.
Assessment
The skill appears coherent and implements only local processing of events. Before installing, review the included handler.ts yourself (or have someone you trust do so) because the package's source/homepage is unknown — even benign-looking code should be vetted. If you plan to record long-term history, note the skill is stateless and you must store entries yourself. As a precaution, test the skill in a sandbox or with non-sensitive example events to confirm it behaves locally and that your OpenClaw runtime enforces no unexpected network access.Like a lobster shell, security has layers — review code before you run it.
latestvk97cyfd3096drm8rq4gdm2j20s83ddbn
License
MIT-0
Free to use, modify, and redistribute. No attribution required.