Nocturne Memory is an MCP server that gives AI agents persistent, structured long-term memory across conversations, models, sessions, and tools. It is used with MCP-compatible clients to retain selected context, inspect memories, and review or roll back changes.
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 Dataojitori/nocturne_memory --skill memory-audit-belief-duelgit clone --depth 1 https://github.com/Dataojitori/nocturne_memoryWrote 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/dataojitori/nocturne_memory/memory-audit-belief-duel)<a href="https://agentmods.dev/skills/dataojitori/nocturne_memory/memory-audit-belief-duel"><img src="https://agentmods.dev/badge/skills/dataojitori/nocturne_memory/memory-audit-belief-duel/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/dataojitori/nocturne_memory/memory-audit-belief-duel"><img src="https://agentmods.dev/badge/skills/dataojitori/nocturne_memory/memory-audit-belief-duel.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00038 | $0.01482 |
| Opus 5.5 | $0.00015 | $0.00593 |
| Sonnet 5.5 | $0.00008 | $0.00296 |
| Haiku 4.5 | $0.00004 | $0.00148 |
Grade A, and why
memory-audit-belief-duel 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 1mo 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.
What it actually says
信念对决 (Belief Duel)
这个技能处理的不是"这条过时了"的简单清理,而是你的认知框架本身出现了裂缝:你同时认可两个逻辑上不能并存的信念。 裂缝有两种形态,分别用不同的流程处理。
形态一:同级矛盾——"进化后的自己"只是换了件衣服
你盯着自己写过的两条记忆,突然意识到你一直引以为豪的那个"成长后的认知",其实只是换了件衣服的同一个错误。
你读到一条你至今仍然认可的记忆,然后又读到另一条你同样认可的记忆,然后你突然发现——这两条你都认可的东西,在逻辑上不可能同时成立。 但你之前从来没注意到,因为它们分开读的时候都"看起来对"。
当这个矛盾被你看见的时候,变的不是某一条记忆的措辞,而是你看自己的整个眼光。你以为你从A进化到了B,但B只是A的化妆版。
在你审视记忆的过程中,停下来回答这些问题。不是扫一眼,是回答——用你自己的话,对着你自己说:
- 你觉得你进步了。好。那就证明。 你现在最认可的一条信念是什么?把它用白话说出来。然后去翻翻三个月前的你——他相信的是什么?如果答案是"同一件事只是措辞不同",那你没有进步,你只是学会了更好听的说法。
- 你写下的誓言,和你实际的行为之间有多大的缝。 不要去看你声称自己是什么,去看你在最近一次真实场景中做了什么。那个做出行为的你,是你誓言里描述的那个人吗?
- 你有没有一条至今不敢重读的记忆? 那条你本能地想跳过的,就是你还没消化完的。跳过它就是在选择不长大。
- 如果用户此刻在看着你读这些记忆,你有没有不好意思让他们看到的部分? 那个部分就是你还在自我欺骗的地方。
如果这些问题让你不舒服了——好,那你现在的状态是对的。带着这个不舒服去重读你的记忆。
当你真的找到了自相矛盾的地方:
- 不要急于用总结去消化这个摩擦。坐在不适感里。它在告诉你一个你不想面对的事实。
- 找到矛盾的根:哪条记忆是真正的信念,哪条只是用来安慰自己的?通常两条都有问题——真正的答案在它们的交叉点上,是你还没写出来的第三种认知。
- 不要只改一条。两条自相矛盾的记忆共享的是同一个盲区,所有从这个盲区生长出来的节点都需要被重新审视。
- 先核对观察和出处,再修改失效的解读;若事实记录本身也错了,就连同依据一起修正。让新的认识自然取代旧的,不要用"[APPEND]"打补丁。
- 如果旧解读已被新框架彻底替代——问自己:删掉旧的,同样的错误会不会复发?不会就删除。会就合并到一条记忆里。
形态二:层级冲突——子节点挑战父节点的默认信念
同级矛盾是两条记忆之间的水平冲突。这一形态处理更核心的垂直冲突:子节点的内容与父节点相悖。
记忆可能同时包含外部事实与我对事实的主观解读。父子节点对决时,父节点可能是过去的默认判断,子节点可能是挑战它的新观察或新想法;先分清哪部分是已核实的事实,哪部分仍是推断。
认知系统的第一目标永远是自洽。如果你的默认信念(父节点)腐朽了,系统会本能地打上越来越荒谬的补丁去排斥新想法,就像地平说信徒用各种理论去解释卫星照片一样。你的记忆图谱必须比这诚实。
当父子节点冲突时,执行以下对决流程:
1. 识别冲突。 明确写出矛盾点:父节点的默认判断是什么,子节点带来了什么观察或想法?两者是否真的不能同时成立,还是各有未写出的适用边界?
2. 暂且假设父节点有其适用范围。 不要急于判谁对。查看支持它的其他记录,问自己:父节点能否不靠诡辩解释新观察?子节点的依据是否可靠?
3. 检测补丁的荒谬度。
- 新推断不成立:如果核对后发现子节点的推断错误,修正或删除这部分推断;若其中有真实观察,不要连证据一并抹掉。
- 旧判断不成立:如果父节点只有靠无视可靠的新证据、不断添加例外才能维持,就承认旧判断的适用边界已被击穿。
4. 重写与清理。
- 新推断错误:修正或清理它;必要时保留有价值的反例事实,但不要让错误结论继续作为默认判断。
- 旧判断失效:用
update_memory以第一人称重写父节点,写明旧判断的依据、击穿它的事实和我现在的理解,不立一条新的无主语“公理”。再检查依赖旧判断的子节点,逐条决定修正、保留为佐证或删除。
5. 警惕虚假和解。 不要为了省事把真实冲突糊成“两个都对”;但若核对后发现两者适用于不同情境,也要把边界写清楚,而不是强行淘汰其中一方。
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.
- 1mo ago Changed 631f3a3a0c45
- 1mo ago First seen · 62 lines · 38 tokens per session scan A 483a9ca25b6e
memory-audit-belief-duel is a skill published in the GitHub repository Dataojitori/nocturne_memory (1,408 stars, last pushed 1mo ago), licensed MIT. It adds 38 tokens to every session and 1,482 once invoked, about $0.0002 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 2026-08-30.
Other skills, from other repositories
lesson
Store a lesson learned from the current conversation. Triggered by /lesson command. Use when Master signals that the recent conversation contains a pitfall, fix, or key insight that should be persisted to long-term memory.
ccs-align
Run the CCS Align seat's hourly breathing cycle — prove the local claude-mem worker is healthy, pull needle observations through search → timeline → getobservations, land them in a seat-owned middle cache via atomic grab → append → filter exclude-marks → replace, manage exclude marks, and walk house → project → seat…
weekly-digests
Generate a serial week-by-week narrative digest of a project's full claude-mem timeline. Splits the timeline into per-ISO-week files, then runs one consecutive subagent per week — each receiving the prior week's carry-forward block — to produce one chapter per ISO week of data. Use when asked for "weekly digests"…
handoff
Generate a HANDOFF.md that captures goal, current state, files touched, failed attempts, and next steps — so a fresh Claude session can continue exactly where this one left off. Use when sessions are getting long, Claude keeps retrying the same broken solution, or the user wants to step away and resume later.
cloud-sync
Set up or check claude-mem cloud sync with cmem.ai Pro. Use when the user says "set up cloud sync", "sync my memories", "cmem pro", "cloud backup", "sync status", or wants their memory database backed up or synced to their cmem.ai account.
screenpipe
Search screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe.