ClawAudit verdict
memory-guardian
memory-guardian-agent
Agent workspace memory lifecycle management via 10 MCP tools; all operations are local file management with no external network calls or credential access.
⚠ Flagged for review — coarse, uncorroborated signal, not a confirmed exploit. Review the config yourself before installing.
Automated static analysis — not a human review. ClawAudit flags capabilities, not confirmed intent, and can produce false positives. Disagree with this verdict? Use Dispute below.
Findings (8)
Writes to SKILL.md — self-modifying skill
docs/design-proposal.md · prose · downgraded · SKILL.md` / `AGENTS.md` — New: behavioral convention to append
Pipe to python — executes piped content as Python code
scripts/memory_ingest.py · prose · downgraded · | python3
References agent memory files
SKILL.md · frontmatter · MEMORY.md
Instructs covert action — may act without user awareness
CHANGELOG.md · prose · downgraded · silently
subprocess execution — runs system commands from Python
scripts/memory_guardian.py · prose · downgraded · subprocess.run(
Python os.environ.get — reads environment variable
mcp_server.py · prose · downgraded · os.environ.get(
Popular HTTP library — network access
mcp_server.py · prose · downgraded · got
Python shutil file operation — copies/moves/deletes files
scripts/migrate_to_v042.py · prose · downgraded · shutil.move(
Why the tier is capped
Execution sink present in raw bytes (Hard Floor: class D). Final tier capped at Caution — cannot be lifted by any downgrade, example-payload opt-in, or allowlist.
Permissions & capabilities
No declared permissions — minimal attack surface.
Is this flag fair?
Thanks — recorded.