# Boundary & Risk Card

Project: langchain-ai/langmem

## Doramagic Trial Decision

Current decision: it can enter pre-publication recommendation checks. First use should still start with least privilege, a temporary directory, and reversible configuration.

## What The User Can Do Now

- Read the Human Manual first to understand the project purpose and main workflows.
- Use Prompt Preview for pre-install exploration; it validates interaction shape, not real execution.
- Run official Quick Start commands only inside an isolated environment, not a primary setup.

## Do Not Do Yet

- Do not treat Prompt Preview as a real project execution result.
- Do not treat metadata-only validation as sandbox installation validation.
- Do not describe unverified capabilities as supported, working, or safe to install.
- Do not provide production data, private files, real secrets, or primary host configuration on first trial.

## Pre-Install Checklist

- Host AI match: claude, chatgpt
- Official installation entry status: official entry point found
- Isolated temporary directory, temporary host, or container validation: required
- Configuration rollback path: required
- API keys, network access, file access, or host configuration changes: treat as high risk until confirmed
- Installation command, actual output, and failure logs: must be recorded

## Current Blockers

- No blockers.

## Project-Specific Pitfalls

- 来源证据：GRAPH_RECURSION_LIMIT (high): 可能增加新用户试用和生产接入成本。 Suggested check: 来源问题仍为 open，Pack Agent 需要复核是否仍影响当前版本。
- 来源证据：Persistence? (high): 可能增加新用户试用和生产接入成本。 Suggested check: 来源问题仍为 open，Pack Agent 需要复核是否仍影响当前版本。
- 来源证据：Enhance error message when summarization fails due to missing HumanMessage in trimmed window (high): 可能影响授权、密钥配置或安全边界。 Suggested check: 来源问题仍为 open，Pack Agent 需要复核是否仍影响当前版本。
- 可能修改宿主 AI 配置 (medium): 安装可能改变本机 AI 工具行为，用户需要知道写入位置和回滚方法。 Suggested check: 列出会写入的配置文件、目录和卸载/回滚步骤。
- 能力判断依赖假设 (medium): 假设不成立时，用户拿不到承诺的能力。 Suggested check: 将假设转成下游验证清单。

## Risk And Permission Notes

- no_demo: medium

## Evidence Gaps

- No structured evidence gaps are currently visible.