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 alibaizhanov/mengram --skill memorygit clone --depth 1 https://github.com/alibaizhanov/mengramWrote 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/alibaizhanov/mengram/memory)<a href="https://agentmods.dev/skills/alibaizhanov/mengram/memory"><img src="https://agentmods.dev/badge/skills/alibaizhanov/mengram/memory/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/alibaizhanov/mengram/memory"><img src="https://agentmods.dev/badge/skills/alibaizhanov/mengram/memory.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 | $0.00080 | $0.00434 |
| Opus 5 | $0.00040 | $0.00217 |
| Sonnet 5 | $0.00016 | $0.00087 |
| Haiku 4.5 | $0.00008 | $0.00043 |
Grade A, and why
Memory recall and capture 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 9d 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
Mengram memory recall and capture
You have access to a persistent memory layer (Mengram) via the bundled MCP server. The memory survives /clear, machine switches, and team handoffs — it's not stored in this conversation.
When to recall
Before answering questions that reference past context the user couldn't have shared in this session, search memory. Examples:
- "What did we decide about the database?"
- "How did we deploy this last time?"
- "What's the project I'm working on?"
- "Did I tell you about my preferences?"
Use the bundled mengram MCP server's search or search_all tools. Default to top-5 results unless the user asks for more.
When to capture
Save proactively when the user shares information worth remembering across sessions:
- Project decisions ("We're going with Postgres because…")
- Preferences ("I always use TypeScript for new projects")
- Constraints ("Production DB pool is capped at 5")
- Known issues ("BM25 search breaks on Chinese queries")
- Workflow steps that completed successfully
Use the mengram MCP server's add or add_text tool. The backend extracts facts, episodes, and procedures automatically — don't pre-structure the input, just pass the relevant conversation text.
When NOT to recall
Don't search for context the user just provided in this turn — that's already in the conversation. Memory is for what's outside the current window.
Cross-tool
The same memory is accessible from Cursor, ChatGPT, Codex, and any other tool with the mengram MCP server configured. When a user says "I told ChatGPT yesterday that…", a memory search will find it.
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.
- 9d ago First seen · 40 lines · 80 tokens per session scan A 1cf645afee21
Memory recall and capture is a skill published in the GitHub repository alibaizhanov/mengram (192 stars, last pushed today), licensed Apache-2.0. It adds 80 tokens to every session and 434 once invoked, about $0.0004 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 2026-08-30.
Other skills, from other repositories
digest-auto
A skill for analysing the current state of an EpisodicRAG system, which retrieves information from records of past events or work sessions.
memory-audit
An entry point for reviewing and maintaining an AI agent's stored memories. It describes how to remove repetition, preserve useful reasoning, and update memories when old conclusions no longer fit.
memory-audit-belief-duel
A guided review process for conflicting beliefs or memories. It examines cases where two conclusions cannot both be true, including conflicts between a general rule and a more specific memory.
memory-audit-discoverability
A review guide for checking whether stored memories can be found at the right time. It focuses on where memories are attached, when they are triggered, whether aliases are missing, and whether a parent has too many children.
memory-audit-pattern-extraction
A method for investigating repeated mistakes by comparing related memories and checking whether an earlier reminder failed. It looks at where the reminder was stored, when it was created, and whether it was strong enough to prevent the mistake.
memory-audit-node-decomposition
A method for splitting an oversized knowledge note into smaller notes, each focused on one independent idea. It also explains how to keep useful core information in the original note.