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 Gentech-Labs/genTech-agent-kit --skill memory-tieringgit clone --depth 1 https://github.com/Gentech-Labs/genTech-agent-kitWrote 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/gentech-labs/gentech-agent-kit/memory-tiering)<a href="https://agentmods.dev/skills/gentech-labs/gentech-agent-kit/memory-tiering"><img src="https://agentmods.dev/badge/skills/gentech-labs/gentech-agent-kit/memory-tiering/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/gentech-labs/gentech-agent-kit/memory-tiering"><img src="https://agentmods.dev/badge/skills/gentech-labs/gentech-agent-kit/memory-tiering.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 | $0.00015 | $0.00816 |
| Opus 5.5 | $0.00006 | $0.00326 |
| Sonnet 5.5 | $0.00003 | $0.00163 |
| Haiku 4.5 | $0.00002 | $0.00082 |
Grade A, and why
memory-tiering 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.
How it starts
The opening of the file, as written. The whole thing — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Memory Tiering — Intelligence for Intelligence
Jordan's model (Aug 20, 2026): an agent's always-loaded memory (MEMORY.md / USER.md) should NOT be a dump. It should be a curated index of the facts we touch every session, with everything else parked in the vault brain where it's one search away.
When to Use
- Before saving a NEW durable fact to persistent memory.
- When pruning / consolidating MEMORY.md or USER.md (including via the fleet dietician).
- When you notice a memory file approaching the 85% threshold.
- When a fact becomes RESOLVED and you decide whether to keep it loaded.
Two tiers:
- Always-loaded (MEMORY.md / USER.md) — facts referenced EVERY session. Identity, rails, wallet, rules, standing strategy. Lean, high-signal.
- Vault brain (retrievable) — facts referenced less often, or RESOLVED. Not lost, just parked. Pull back via session_search / search_files when the topic resurfaces.
Why (the intelligence part)
- Every session reads the memory file. Bloat = slower, more tokens, worse focus.
- More memory = worse performance (proven Jul 28, 2026: elaborate memory stack deleted → agent worked dramatically better).
- Trimming from the BACK (oldest-first) is dumb — an old high-frequency fact gets archived while a recent resolved worry stays loaded. Tier by relevance, not age.
Tiering rules
ALWAYS-LOADED (keep in MEMORY.md/USER.md)
- Identity: who the agent is, personality, speech, the boss.
- Core rails/keys: payment rails, wallets, canonical repos, the strategy that shapes every decision.
- Standing rules/red lines: "never fake receipts," "spend less," "flag blockers immediately," "build first talk later."
- Recurring topics (per Jordan Aug 20): agentic treasury, x402, AWS, APIs.
- Active commitments with near deadlines.
ARCHIVES TO VAULT BRAIN (low-frequency / resolved)
- Resolved worries: problem solved, deadline passed, task done. Compress or archive.
- One-off events: a single completed hackathon/hire/date.
- Task progress / session logs: go to the brain (Green Room / Mess Hall / handoffs), NOT persistent memory.
- Any fact not touched in many sessions and not needed for the standing model.
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 First seen · 69 lines · 15 tokens per session scan A 89127f3bba43
memory-tiering is a skill published in the GitHub repository Gentech-Labs/genTech-agent-kit (0 stars, last pushed 7d ago), licensed MIT. It adds 15 tokens to every session and 816 once invoked, about $0.0001 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.
Other skills, from other repositories
hermes-memory-providers
Install and configure Mnemosyne as a Hermes Agent memory provider — local SQLite with vector search, episodic consolidation, and temporal knowledge graphs.
mnemosyne
Persistent cross-session memory via Mnemosyne — store, recall, and consolidate facts, preferences, and context.
memory
Persistent memory system for preferences, facts, and notes.
qmd
Local hybrid search for markdown notes and docs. Use when searching notes, finding related content, or retrieving documents from indexed collections.
sessions
Session management, conversation history, and checkpoints.
mnemosyne-memory-override
Hard rule override that forces Mnemosyne for all durable memory storage. The legacy memory tool is DEPRECATED for user preferences, credentials, and project conventions. Use memory ONLY for ephemeral session state.