Back to skill
Skillv0.2.0
ClawScan security
Memtrap Skill · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
SuspiciousApr 22, 2026, 1:03 PM
- Verdict
- suspicious
- Confidence
- medium
- Model
- gpt-5-mini
- Summary
- The skill's stated purpose (testing/hardening agent memory) matches the operations described in SKILL.md, but there are inconsistent registry metadata and packaging signals plus an optional public submission feature that could expose memory—inspect the package/source before using on real secrets.
- Guidance
- This skill appears to do what it says (Python package that benchmarks and wraps agent memory), but there are a few red flags: the registry metadata omitted the install spec while SKILL.md requires pip/python3; the SKILL.md offers a public 'submit' which could leak memory/context; and no package/source files are included in the bundle so you must trust the external pip package. Before installing or using on real agents: 1) verify the memtrap package on PyPI and inspect its GitHub repo/source code (https://github.com/shaymizuno/memtrap is listed in SKILL.md) for surprising network calls or telemetry; 2) never run benchmarks or call memtrap submit on production memory or any data containing secrets—use sanitized or synthetic contexts in a sandbox; 3) prefer installing in an isolated environment (virtualenv/container) and review the package's maintainers, release history, and license; 4) if you need higher assurance, request the upstream source tarball and audit it or run it in an offline environment. If the upstream project and package provenance check out and you avoid submitting sensitive contexts, the tool can be useful; otherwise treat it as untrusted code.
Review Dimensions
- Purpose & Capability
- noteSKILL.md describes a Python package (memtrap) and shows Python APIs/CLI for benchmarking and wrapping agent memory; that matches the stated purpose. However, the registry metadata claims no install spec/no required binaries while SKILL.md declares a pip install and dependency on python3—this metadata mismatch is unexpected and should be resolved.
- Instruction Scope
- concernInstructions focus on running MemTrap in benchmark or active (wrap_memory) modes, and providing a CLI submit command to publish results. Wrapping an agent's memory and running benchmark code is coherent, but the 'memtrap submit' command posts data to a public leaderboard (potentially sensitive memory/context). The doc also claims 'Zero telemetry' while offering a public submit flow—this contradiction is a risk if users submit real memory content or secrets.
- Install Mechanism
- concernInstallation is via pip (memtrap). Pip installs are common but pull remote code that will run locally; SKILL.md includes an install block but the registry metadata omitted it — this inconsistency reduces trust. Because there is no bundled code in the skill bundle and the package source is not validated here, you should verify the PyPI package and upstream GitHub repository before installing.
- Credentials
- okThe skill does not request environment variables, credentials, or config paths in the registry. The SKILL.md's runtime examples operate on in-memory agent objects and do not ask for unrelated credentials.
- Persistence & Privilege
- okThe skill is not marked 'always: true' and does not request elevated or persistent platform privileges in the provided instructions. Its active mode modifies only the agent's memory object (wrap_memory) which is in-scope for a memory-hardening tool.