Model Usage Linux

AdvisoryAudited by Static analysis on Apr 30, 2026.

Overview

No suspicious patterns detected.

Findings (0)

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.

NoteHigh Confidence

ASI06: Memory and Context Poisoning

What this means

Running the skill lets it process local OpenClaw session logs to calculate usage totals.

Why it was flagged

The script reads local OpenClaw session JSONL files, which may contain sensitive conversation history even though the script only summarizes usage fields.

Skill content

DEFAULT_SESSIONS_DIR = os.path.expanduser("~/.openclaw/agents/main/sessions") ... jsonl_files = list(sessions_dir.glob("*.jsonl"))

Recommendation

Use it only on session directories you are comfortable processing, and prefer --sessions-dir pointing to a limited directory if you want narrower scope.

NoteHigh Confidence

ASI02: Tool Misuse and Exploitation

What this means

A user expecting --session to limit analysis to one session could still get aggregate results from all sessions in the directory.

Why it was flagged

The command-line help suggests a specific-session scope, but main() never uses args.session and instead iterates over every *.jsonl file in the chosen sessions directory.

Skill content

p.add_argument("--session", help="Specific session ID or 'all' (default: all)") ... for f in jsonl_files:

Recommendation

Do not rely on --session for scoping until the script implements it; use a separate directory with only the intended session files if precise limiting is needed.