my-test
Security checks across static analysis, malware telemetry, and agentic risk
Overview
The code does not show exfiltration or destructive behavior, but the package identity is inconsistent and the skill encourages persistent agent-memory changes that can affect future sessions.
Install only if you want persistent self-improvement memory for your agent. Verify the package identity and source first, avoid following install commands for a different package unless intentional, keep learnings free of secrets, and require manual review before promoting notes into AGENTS.md, SOUL.md, TOOLS.md, CLAUDE.md, MEMORY.md, or Copilot instruction files.
Static analysis
No static analysis findings were reported for this release.
VirusTotal
VirusTotal findings are pending for this skill version.
Risk analysis
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.
A user may think they are installing one package but be guided toward a different package or repository.
The registry metadata supplied for review identifies this package as my-test version 1.0.1, while embedded metadata and SKILL.md setup text point to a different skill identity. That makes provenance and installation target ambiguous.
"slug": "self-improving-agent", "version": "1.0.11"
Verify the publisher and intended source before installing. The package should align registry name, embedded metadata, and installation instructions.
Incorrect, sensitive, or adversarially influenced notes could be reused later and change how the agent behaves across sessions or projects.
The skill asks agents to persist learnings and promote them into files that influence future sessions, but the artifact does not clearly require user review, privacy filtering, retention limits, or rollback before promotion.
OpenClaw injects these files into every session: AGENTS.md, SOUL.md, TOOLS.md, MEMORY.md ... Broadly applicable learning | Promote to CLAUDE.md, AGENTS.md, and/or .github/copilot-instructions.md
Keep routine logs separate from persistent instruction files, require explicit user approval before promotion, and periodically review or remove stored learnings.
If used carelessly, learnings or transcript details from one session could be exposed to another session or background agent.
The docs describe cross-session transcript access, message sending, and background sub-agent spawning. This is purpose-aligned for sharing learnings, but the artifact does not describe privacy boundaries for transcripts or handoffs.
sessions_history — Read transcript from another session ... sessions_send — Send message to another session ... sessions_spawn — Spawn a background sub-agent
Only share non-sensitive summaries across sessions and confirm which session or sub-agent should receive the information.
Future sessions may automatically receive self-improvement reminders, which can influence the agent to keep logging and promoting learnings.
When the optional hook is enabled, it injects a reminder at agent bootstrap. The code appears scoped and does not show network access or file modification, but it is persistent behavior across sessions.
event.type !== 'agent' || event.action !== 'bootstrap' ... event.context.bootstrapFiles.push({ path: 'SELF_IMPROVEMENT_REMINDER.md', content: REMINDER_CONTENT, virtual: true })Enable the hook only if you want this persistent reminder behavior, and disable it if it becomes intrusive or inappropriate.