Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add gaooooosh/limem-agent-plugin --skill limem.remembergit clone --depth 1 https://github.com/gaooooosh/limem-agent-pluginWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/gaooooosh/limem-agent-plugin/limem.remember)<a href="https://agentmods.dev/skills/gaooooosh/limem-agent-plugin/limem.remember"><img src="https://agentmods.dev/badge/skills/gaooooosh/limem-agent-plugin/limem.remember/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/gaooooosh/limem-agent-plugin/limem.remember"><img src="https://agentmods.dev/badge/skills/gaooooosh/limem-agent-plugin/limem.remember.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00125 | $0.01671 |
| Opus 5.5 | $0.00050 | $0.00668 |
| Sonnet 5.5 | $0.00025 | $0.00334 |
| Haiku 4.5 | $0.00013 | $0.00167 |
Grade A, and why
limem.remember scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 139 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/limem.remember — 把用户陈述固化为长期记忆(event 写入 + mention 抽取)
何时调用
- 用户显式说 "remember X" / "以后这个项目不要 Y" / "always Z" / "记住 W"
- 用户对刚才输出强烈纠正("不对,应该…"),此时主动建议保存为 feedback
- 用户在 Claude Code 输入
/limem.remember <text>或 Codex 输入$limem.remember <text>时
边界(v3)
本 skill 只写 event:
- event:进入 BM25 软召回池,summary 由后端 LLM 抽取,importance 由 mem_type 决定。
- mentions(旧字段名
entities):抽取的 canonical / aliases 进入 BM25 tag-token 与本地raw_metadata.canonicals,召回时增强匹配;不再注册为后端 entity。 principal_ids:根据 scope / mem_type 自动推断本次 event 应挂的 principals (user / agent / project),写入本地 metadata,soft 召回时降权过滤。
本 skill 不写 principal 的 markdown 档案。要附长文档(用法 / 反例 / 约定)请用
/limem.pattern project|user|agent——这是 v3 的设计分工。
LiMem 现在按上下文相关性召回,canonical / aliases 会作为 trigger 辅助命中;它不再适合保存"必须每轮生效"的无条件约束。始终生效的要求应写入 CLAUDE.md、AGENTS.md 或全局系统指令。
处理步骤(严格按顺序)
Step 1 — 解析输入与 mention 抽取
把 $1 / $ARGUMENTS 当成"用户原话"。从中抽出关键 mentions(每个抽这些字段):
canonical:mention 的规范名(如npm run dev、docker rebuild、react-query、/api/v2)role:在规则中扮演的角色:forbidden— 被禁止 / 不允许的事preferred— 被推荐 / 应该用的事subject— 规则关心的主题但本身不禁不推(如"前端"、"数据库迁移")neutral— 纯 fact 类引用
aliases(可选,建议 2–6 个):跨工具家族、跨语言、口语 / 正式形态的同义说法。 这些进入 BM25 tag-token 与本地 metadata 镜像,是匹配口语化 prompt 的关键。description(可选):一句话描述该 mention 在本规则下的含义(≤ 60 字符),仅落本地 metadata 不上行后端。
抽取要点:
- 单条
remember通常 1–3 个 mention;不要无中生有。 - canonical 必须是用户原话或一对一可还原的规范化;不要意译。
- aliases 给跨工具家族(npm/yarn/pnpm/bun)、跨语言(中文/英文)、口语 / 正式形态。
- 不要枚举大量触发短语——长文档与"档案级"约定走
/limem.pattern。
Few-shot 范例
例 1:"以后这个项目不要用 npm run dev,直接 docker rebuild"
{
"text": "以后这个项目不要用 npm run dev,直接 docker rebuild",
"scope": "project",
"mem_type": "rule",
"importance": 0.9,
"entities": [
{
"canonical": "npm run dev",
"role": "forbidden",
"aliases": ["npm dev", "yarn dev", "pnpm dev", "bun dev", "起一下 dev"]
},
{
"canonical": "docker rebuild",
"role": "preferred",
"aliases": ["docker compose up --build", "重建 docker", "重新构建容器"]
}
]
}
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2mo ago First seen · 139 lines · 125 tokens per session scan A a1fb4e2a9ecb
limem.remember is a skill published in the GitHub repository gaooooosh/limem-agent-plugin (2 stars, last pushed 5mo ago), licensed MIT. It adds 125 tokens to every session and 1,671 once invoked, about $0.0005 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
limem.remember compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| limem.remembergaooooosh/limem-agent-plugin | Skill | 2 | 5mo ago | MIT | A | 125 | Claude Code |
| knowledge-ingestopenclaw/openclaw | Skill | 394k | today | MIT | A | 80 | Claude Code, Codex |
| mem0-oss-to-platformmem0ai/mem0 | Skill | 69k | yesterday | Apache-2.0 | A | 217 | Claude Code, Codex |
| forgetasgeirtj/system_prompts_leaks | Skill | 71k | yesterday | CC0-1.0 | A | 0 | Claude Code, Codex |
| Cortexdanielmiessler/LifeOS | Skill | 20k | 2d ago | MIT | A | 184 | Claude Code |
| ha-data-storesshiwenwen/hope-agent | Skill | 1.9k | today | MIT | A | 115 | Claude Code, Codex |
| memoryopensquilla/opensquilla | Skill | 7.1k | 7d ago | Apache-2.0 | A | 44 | Claude Code, Codex |