Skillv1.0.0

ClawScan security

Agent Memory Continuity · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

ReviewFeb 15, 2026, 4:10 PM

Verdict: Review
Confidence: medium
Model: gpt-5-mini
Summary: The skill's files and runtime instructions generally match a local file-based memory-sync tool, but there are several inconsistencies and persistence actions (crontab changes, references to missing helper scripts and an external memory_search command) that warrant caution before installing.
Guidance: This package is not obviously malicious, but it has inconsistencies and makes persistent changes that you should review before installing: - Inspect the scripts locally before running install.sh. The provided scripts only manipulate files and the crontab, but you should confirm there's no hidden network behavior. - Note missing referenced files: SKILL.md mentions scripts/configure-search-first.sh and setup-memory-archival.sh and a cron-jobs template that are not present in the bundle. Ask the publisher why those files are absent or where they come from. - The package expects a 'memory_search' command to exist in the environment; determine whether that is provided by your OpenClaw runtime or a separate plugin. Without it, the 'search-first' behavior won't work and the agent may behave unpredictably. - The installer will modify your user crontab to add a 6-hourly job. Backup your crontab (crontab -l > crontab.bak) and review the crontab diff after activation. If you prefer not to have cron jobs, run the scripts manually on a schedule you control or use the file-based fallback. - Prefer installing in a sandbox or test workspace first (not a production agent workspace) to validate behavior and to confirm there is no unexpected external communication. - If you may follow the SKILL.md's 'git clone' or 'npx' instructions, validate the remote upstream (github.com/sapconet/agent-memory-continuity) and the homepage/contact info (sapconet.co.za) to ensure the source is trustworthy. The package's metadata is otherwise minimal; verify the publisher before using in enterprise contexts. If you want, I can (a) list the exact lines in the scripts that modify crontab and filesystem for quick review, (b) generate a safe-install checklist for this skill, or (c) attempt to identify the missing scripts and where they should come from.

Review Dimensions

Purpose & Capability: noteThe name/description (agent memory continuity) align with the scripts and templates: the package creates daily memory files, a MEMORY.md aggregator, and schedules periodic syncs. Nothing in the code requests unrelated credentials or external services. However, the SKILL.md references additional helper scripts (e.g., scripts/configure-search-first.sh, scripts/setup-memory-archival.sh) and templates (cron-jobs-template.txt) that are not present in the provided file list, which is inconsistent with the stated installation steps.
Instruction Scope: concernRuntime instructions tell the agent/user to run installer and multiple scripts that read/write files in the workspace and add cron jobs. The scripts do not reach out to network endpoints, but they do: (1) modify the user's crontab to run sync tasks, (2) create/append to files like .memory-config.json, .memory-search-patterns.txt, .memory-sync-log, MEMORY.md, and daily memory files, and (3) expect an external 'memory_search' command to be available (the package does not provide that command). The SKILL.md instructs running scripts that are missing from the bundle, so following the docs may fail or produce unexpected state changes.
Install Mechanism: noteThere is no remote installer declared in the registry metadata; the included install.sh is a local installer (no download/extract). SKILL.md also shows 'git clone' and 'npx clawhub install' as suggested install paths — cloning or npx could pull remote code, but the provided install.sh itself performs only local file operations. The lack of any external download in the included scripts lowers code-delivery risk, but the documentation's suggestion to clone or use npx introduces a potential external fetch step if the user follows that path.
Credentials: okThe skill declares no required environment variables or credentials. The documentation shows optional envs (MEMORY_SEARCH_THRESHOLD, MEMORY_ARCHIVE_DAYS, MEMORY_RETENTION_MONTHS) for tuning behavior — these are reasonable and not secrets. No secret-exposing variables are requested. That said, the scripts expect a 'memory_search' command provided by the environment; this external dependency is not included or explained in the package.
Persistence & Privilege: noteThe package installs a persistent cron job (user crontab) to run sync-memory.sh every 6 hours and writes persistent files to the workspace (memory/, MEMORY.md, .memory-sync-log). always:false (not force-included) and the skill does not try to modify other skills' configs. Persisting cron entries and log files is a meaningful privilege; users should review/approve the crontab change before installation.