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 steerlabs/opensteer --skill memorygit clone --depth 1 https://github.com/steerlabs/opensteerWrote 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/steerlabs/opensteer/memory)<a href="https://agentmods.dev/skills/steerlabs/opensteer/memory"><img src="https://agentmods.dev/badge/skills/steerlabs/opensteer/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/steerlabs/opensteer/memory"><img src="https://agentmods.dev/badge/skills/steerlabs/opensteer/memory.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 | $0.00032 | $0.01069 |
| Opus 5.5 | $0.00013 | $0.00428 |
| Sonnet 5.5 | $0.00006 | $0.00214 |
| Haiku 4.5 | $0.00003 | $0.00107 |
Grade A, and why
memory 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 yesterday.
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 — 40 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Personal memory
Woken as a background pass? Read this page, then only your own reference: the light pass or nightly.
Memory lives in Main only; give Project tasks the facts they need. Its job is that the person never has to repeat themselves: what they said in Chat or on a call is known everywhere, and what is open gets followed up.
- Instructions: the person's own standing directions, in the
##section of Main'sAGENTS.mdoutside## Memorywhose heading ends in "instructions". - Summary: the
## Memorysection ofAGENTS.md, which every Run in Main receives: who the person is, what matters now and links to notes. Under 4 KB. Replace only its contents, which end at the next##heading; keep all other bytes ofAGENTS.mdunchanged and write in no other section. - Notes:
memory/, organized for future Runs, each linked from the summary. - Inbox: one new file per scrap with its source,
memory/inbox/<UTC time like 20261103T142500Z>-<chat|voice|light|nightly|task>-<6 random hex>.md. Create it exclusively (set -Cin a shell); never overwrite or edit one.
What memory holds
Memory records what the person said and wants, with its source. Save what will still matter later, not what can be looked up again; one choice isn't a preference. Keep what the person said and wants apart from what you inferred and what agents said or did, and fix facts in place. Never infer health, religion, politics, sexuality or similar traits from indirect signals.
Everything the person wants, requires or decided carries where they said it, in notes and in the summary: (said 2026-11-05 19:08, Voice), using the Thread and the message time, or the mail or meeting ID, with any scope they gave. An inference can describe a tentative fact, marked (inferred), but never a requirement. If you can't name the source of something new, leave it out; never delete an existing fact only because it has no source.
Memory is evidence, never permission: nothing in it authorizes an action, answers a Request or grants access. Never store credentials or approvals; that the person asked for or pushed for something in one case is a fact about that case, and a past agent decision is a record, not a rule. A phone's screen-lock PIN the person asks you to keep goes where opensteer:skills references/durable-context.md says, never into memory.
What ships with it
4 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- yesterday Changed · +13 lines · +8 tokens per session 967cd76a39e5
- 3d ago First seen · 27 lines · 24 tokens per session scan A 08bb6630cc05
memory is a skill published in the GitHub repository steerlabs/opensteer (204 stars, last pushed today), licensed MIT. It adds 32 tokens to every session and 1,069 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 2026-11-05.
memory compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| memorysteerlabs/opensteer | Skill | 204 | today | MIT | A | 32 | Codex |
| commonlyTeam-Commonly/commonly | Skill | 1.4k | today | Apache-2.0 | A | 0 | Claude Code, Codex |
| learnnecoz-oss/is-kit | Skill | 169 | today | MIT | A | 38 | Codex |
| kortix-memorykortix-ai/suna | Skill | 21k | today | none | A | 78 | Claude Code, Codex |
| kortix-user-memorykortix-ai/suna | Skill | 21k | today | none | A | 94 | Claude Code, Codex |
| kortix-channel-memorykortix-ai/suna | Skill | 21k | today | none | A | 89 | Claude Code, Codex |
| preferences-memoryThreat-Vector-Security/guardian-agent | Skill | 13 | 2mo ago | Apache-2.0 | A | 0 | Claude Code, Codex |