MH 1password

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.

What this means

After authorization, the agent could access sensitive credentials available to the signed-in 1Password account if the task or later instructions are too broad.

Why it was flagged

The skill is explicitly designed to authenticate to 1Password and read or use vault secrets. That is purpose-aligned, but the artifacts do not clearly restrict which account, vault, item, or secret the agent may access.

Skill content
description: Set up and use 1Password CLI (op)... reading/injecting/running secrets via op... Verify access inside tmux: `op whoami` (must succeed before any secret read).
Recommendation

Use this only with explicit user direction. Require confirmation for each `op read`, `op run`, `op inject`, account, vault, item, and output destination; prefer least-privilege or test vault access.

What this means

A secret could be exposed in terminal output, agent-visible logs, chat, or local files if these examples are used carelessly.

Why it was flagged

The examples document raw op commands that can print secrets, disable masking, or write secrets to local files. They are not auto-run, but they are high-impact patterns without accompanying approval or destination checks.

Skill content
`op read --out-file ./key.pem op://app-prod/server/ssh/key.pem`; `op run --no-masking -- printenv DB_PASSWORD`; `op inject -i config.yml.tpl -o config.yml`
Recommendation

Avoid `--no-masking` and printing secrets. Ask before writing any secret file, verify paths, and prefer masked or process-local secret injection.

ConcernMedium Confidence
ASI06: Memory and Context Poisoning
What this means

Secrets printed during a tmux session could be copied into the agent conversation or retained in logs/context.

Why it was flagged

The skill relies on capturing tmux pane output for op commands and also warns not to paste secrets into logs or chat. If secret-producing commands are captured, sensitive values may enter the agent context despite the guardrail.

Skill content
`tmux -S "$SOCKET" capture-pane -p -J -t "$SESSION":0.0 -S -200` and `Never paste secrets into logs, chat, or code.`
Recommendation

Do not capture panes after secret-producing commands. Keep secret values out of chat, logs, summaries, and saved context; use commands that avoid printing secret material.