1

    Agent Memory Privacy Check

    Audit AI agent memory files for privacy risk and bloat.

    Updated May 2026
    0 installs

    Free

    One-time purchase

    Included in download

    • Downloadable skill package
    • Works with Codex CLI, Cursor
    • Instant install

    Sample Output

    A real example of what this skill produces.

    Status: CONCERN Must Remove: - API key found in line 14 (sk-proj-...) - User's home address in USER.md Compress: - Project 'Alpha' logs from 2023 (800 chars) Suggested Replacement for MEMORY.md: "User prefers Python 3.11 for ML tasks. Currently migrating Project Alpha to FastAPI."

    About This Skill

    A quick privacy check for agent memory

    Agent Memory Privacy Check audits durable AI memory files such as SOUL.md, MEMORY.md, USER.md, AGENT.md, USER_PROFILE.md, WORKING_MEMORY.md, HANDOFF.md, and OBSERVATIONS.md.

    What it finds

    It flags secrets, stale assumptions, unsupported personal claims, raw transcript hoarding, sensitive personal data, and memory bloat that can make an agent less useful or less safe.

    Hermes-aware

    For Hermes-style memory, it checks the known hot-memory budgets: MEMORY.md at 2,200 characters and USER.md at 1,375 characters.

    Trust builder

    This free skill is a lightweight audit. It helps users see the value of clean memory before moving to a deeper continuity-curation workflow.

    Use Cases

    • Scan memory files for leaked API keys, tokens, and personal PII.
    • Enforce Hermes-style character budgets to prevent context window saturation.
    • Identify and prune stale project data and redundant assistant observations.
    • Sanitize agent memory profiles before sharing or selling them.
    • Differentiate between durable user preferences and temporary task fluff.

    Reviews

    No reviews yet - be the first to share your experience.

    Only users who have downloaded or purchased this skill can leave a review.

    Security Scanned

    Passed automated security review

    Permissions

    No special permissions declared or detected

    No runtime dependencies. Works with SKILL.md-compatible agents including Claude Code, Codex CLI, Cursor, VS Code Copilot, Gemini CLI, and OpenClaw. Do not paste secrets or private personal data unless you intend to have it identified and removed.

    Frequently Asked Questions

    Free