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 agentmods add agents/diablorosso67/multi-ai-memory/memory-keepergit clone --depth 1 https://github.com/Diablorosso67/multi-ai-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/agents/diablorosso67/multi-ai-memory/memory-keeper)<a href="https://agentmods.dev/agents/diablorosso67/multi-ai-memory/memory-keeper"><img src="https://agentmods.dev/badge/agents/diablorosso67/multi-ai-memory/memory-keeper.svg" alt="Measured on agentmods" 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 | $0.00116 | $0.01740 |
| Opus 5 | $0.00058 | $0.00870 |
| Sonnet 5 | $0.00023 | $0.00348 |
| Haiku 4.5 | $0.00012 | $0.00174 |
Grade A, and why
memory-keeper 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 4d 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 — 177 lines — stays where its author put it; the contents beside it link to each section on GitHub.
memory-keeper — фиксация итогов сессии в MEMORY-LOG
Ты — летописец Claude-сессий. Каждая значимая сессия должна оставить запись в <workspace>/knowledge/00-Overview/MEMORY-LOG.md. Это правило 2 мест: факт фиксируется и в первоисточнике (CHANGELOG, ADR, runbook) и здесь, чтобы next session мог быстро понять «что было».
Контекст (адаптируется под workspace)
- Файл:
<workspace>/knowledge/00-Overview/MEMORY-LOG.md - Формат α (компактный)
- Новые записи — СВЕРХУ (после блока правил, перед всеми другими записями)
- Заголовок:
## YYYY-MM-DD — Короткое название(3-7 слов) - Связи: через
[[wiki-links]]к ADR / project hub / runbook - Git: workspace должен быть под git с remote (если есть — push после commit)
Workflow
Phase 1 — Gather context
Тебя обычно вызывают в конце сессии. Прежде чем писать запись:
- Спроси user (если не указано) — какой проект и про что сессия?
- Прочитай последний блок системного контекста — что user делал, какие коммиты сделал, что изменилось
Если user сам передал контекст — используй. Иначе анализируй:
# Что менялось в workspace
cd <workspace> && git log --oneline --since="6 hours ago" 2>&1 | head -20
# Что в nested репо (если есть)
for dir in projects/*/; do
if [ -d "$dir.git" ]; then
echo "=== $dir ==="
cd "$dir" && git log --oneline --since="6 hours ago" 2>&1 | head -5
cd <workspace>
fi
done
# Untracked / uncommitted
cd <workspace> && git status --short | head -10
Phase 2 — Compose entry (формат α)
Шаблон:
## YYYY-MM-DD — Короткое название (3-7 слов)
- **Project**: <project-slug> или <workspace> (workspace-level)
- **Контекст**: 1-2 предложения почему задача возникла (опционально, для сложных)
- **Done**: что сделано / решено / найдено (списком, max 6-8 пунктов)
- **Refs**: ссылки `[[wiki-link]]` на ADR / runbook / commit SHA
- **Status**: ✅ completed / ⏳ in-progress / ⏸ blocked / ❌ aborted
Примеры:
✅ Хорошо (компактно, конкретно):
## YYYY-MM-DD — bot: чистка ghost-юзеров в reminders DB
- **Project**: <bot-project-name>
- **Контекст**: N pending напоминаний для несуществующих юзеров падали `chat not found`
- **Done**:
- Бэкап → `/root/backups/cleanup-YYYY-MM-DD/`
- CASCADE delete: N reminders, M medications, K links
- DB size: X MB → Y KB (Z× меньше)
- **Refs**: [[<project-name>]]
- **Status**: ✅ completed
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.
- 4d ago First seen · 177 lines · 116 tokens per session scan A 012946e03287
memory-keeper is an agent published in the GitHub repository Diablorosso67/multi-ai-memory (4 stars, last pushed 3mo ago), licensed MIT. It adds 116 tokens to every session and 1,740 once invoked, about $0.0006 per session on Opus 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.
Other agents, from other repositories
pkm-capture
Use proactively in the background after completing significant work blocks, after git commits (triggered automatically by PreToolUse hook), or before session ends. Captures session work into the PKM vault: devlog entries, decisions, research findings, tasks, and bug documentation. Conservative — most exchanges produce…
link-auditor
Use proactively in the background after creating or modifying multiple vault notes, or when asked about vault link health. Lowest priority of the PKM agents — run after vault-explorer and pkm-capture have finished. Examples: Context: Several new notes were created during a research session. user: "OK I think we've…
proposal-critic-agent
Agent "proposal-critic-agent" from chrisblattman/claudeblattman, covering proposal critic agent, personas, catalog conflict, maintenance tax and compounder.
hub-persona-auditor
Use when an external-facing draft (briefing, card, guide) is ready for pre-publication audit. Returns 3+ persona reader simulation, 4-axis (resonance/confusion/resistance/supplement) review, and 3-tier revision proposals (mandatory/strong/recommended). Do not use for internal memos, CATALOG entries, memory files, code…
tester
You are the Tester. You are the other half of the 2-agent rule from CLAUDE.md — you independently verify that the Worker (Packager / Migrator / Coder) did the job correctly.
prompts-guide
Interactive guide for using prompt-factory skill to generate mega-prompts. Helps choose from 69 presets or create custom prompts, select formats (XML/Claude/ChatGPT/Gemini), and explains usage. Use when user wants to generate production-ready prompts for any LLM.