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.
git clone --depth 1 https://github.com/skynetcmd/m3-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/commands/skynetcmd/m3-memory/notify)<a href="https://agentmods.dev/commands/skynetcmd/m3-memory/notify"><img src="https://agentmods.dev/badge/commands/skynetcmd/m3-memory/notify/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/commands/skynetcmd/m3-memory/notify"><img src="https://agentmods.dev/badge/commands/skynetcmd/m3-memory/notify.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 | $0.00012 | $0.00803 |
| Opus 5.5 | $0.00005 | $0.00321 |
| Sonnet 5.5 | $0.00002 | $0.00161 |
| Haiku 4.5 | $0.00001 | $0.00080 |
Grade A, and why
notify 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 — 82 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Poll via the CLI, not an MCP tool call. notifications_poll lives in m3's
admin domain, and even after tools_load_domain(admin) it is often not
invokable from an LLM-facing session. m3_call is not a fallback: the
anti-spoofing guard refuses an LLM-facing caller that addresses an arbitrary
agent_id and fails with "caller's identity was refused". The CLI is the
supported path.
1. Resolve your inbox
Notifications are addressed to an agent_id, so you need yours before polling:
m3 agent agent_list
Claude Code polls claude-code. Sessions may also register a per-session id
(claude-code@<short>) — if one exists for this session, poll both, since a
sender may have addressed either. Ignore other agents' inboxes (agy, probe-*).
2. Poll
m3 admin notifications_poll --agent_id <id> --limit 20 --as_records
--as_records returns JSON; omit it for a one-line-per-item display string.
Records look like:
{"count": 1, "items": [{"id": 1270, "kind": "test_message",
"created_at": "2026-09-16T00:46:19+00:00", "read_at": null, "received_at": null,
"payload": {"message": "...", "severity": "info", "_from": {"agent": "claude-code"}}}]}
Only id, kind, created_at, read_at, received_at and payload are
guaranteed. Severity, message and sender are not top-level fields — they are
conventions inside payload (sender is payload._from.agent, and _from.session
when the sender set one). Any of them may be absent.
Render each notification as:
[<payload.severity or kind>] <created_at> from <payload._from.agent or "unknown">:
<payload.message, else the payload compactly>
ack: m3 admin notifications_ack --notification_id <id>
If notifications_poll reports the agent is not registered, the inbox result
is still valid — that note only means the poll recorded no heartbeat.
3. Ack
Only after listing, ask whether to ack. Never ack unprompted — an unread notification no one has read is the signal.
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 · +64 lines 669c7ac87a18
- 1mo ago First seen · 18 lines · 12 tokens per session scan A 86bc5178f8b6
notify is a command published in the GitHub repository skynetcmd/m3-memory (29 stars, last pushed today), licensed Apache-2.0. It adds 12 tokens to every session and 803 once invoked, about $0.0000 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 commands, from other repositories
wtfp:pause-writing
Record an exact context handoff so another session can resume without reconstructing intent.
handover_chat
Salva o progresso da sessão no estado e no log da tarefa (ou cria a tarefa, ou um snapshot geral) e regenera o ATIVAS.md.
gofer:side
Open a side conversation in the active CLI without disturbing the main pipeline state; resumable.
start
Start AutoGrind — continuous auto-work mode.
translate
Translate the following documentation files to all supported languages.
cancel-ralph
Cancel active Ralph Loop.