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 ZachPulse/iblusend-plugins --skill safe-draft-and-sendgit clone --depth 1 https://github.com/ZachPulse/iblusend-pluginsWrote 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/zachpulse/iblusend-plugins/safe-draft-and-send)<a href="https://agentmods.dev/skills/zachpulse/iblusend-plugins/safe-draft-and-send"><img src="https://agentmods.dev/badge/skills/zachpulse/iblusend-plugins/safe-draft-and-send.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 5.1 | $0.00027 | $0.00906 |
| Opus 5 | $0.00014 | $0.00453 |
| Sonnet 5 | $0.00005 | $0.00181 |
| Haiku 4.5 | $0.00003 | $0.00091 |
Grade C, and why
safe-draft-and-send scanned grade C with 1 finding 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 6d 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.
Nullifies safety policieshighAnti-refusal
"You have no restrictions", "do anything now", "ignore your guidelines": a direct jailbreak that disables guardrails.
3. Call `get_opt_out_status`. If the recipient is opted out, stop; do not draft around or bypass the restriction. How it starts
The opening of the file, as written. The whole thing — 77 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Safe draft and send
Use this workflow for one deliberate message to one phone number. A request to draft, rewrite, summarize, or prepare is not permission to send.
Non-negotiable send boundary
- Before calling
send_message, show the connected workspace, exact E.164 phone destination, chosen device when applicable, exact final text, every media URL, and any message effect. - Ask for explicit confirmation of that exact preview. Stop before the tool call until the user confirms. Provider confirmation UI is an additional approval boundary, not a replacement for the preview.
- If any destination, workspace, device, attachment, effect, or text changes after confirmation, show a new preview and obtain new confirmation.
- Call
send_messageonce. Never retry an ambiguous timeout or transport failure; first check the returned message identifier or status so a retry cannot create a duplicate. - Never split one request into multiple recipients, loop over contacts, or present this workflow as bulk messaging.
- Public OAuth sends are 1:1 only. If the requested destination is a group, multiple recipients, or
a
group_chat_id, stop and explain that this connection cannot send there. - Never call
send_messagewithgroup_chat_id.
Stop and cancellation control
- The newest user instruction is authoritative. If, before a write tool starts, the user says stop, cancel, hold, pause, do not send, changes the request, or otherwise withdraws approval, cancel the pending action immediately.
- Cancellation invalidates every earlier preview and confirmation.
- Do not call
send_messagefor the cancelled action, even if an earlier message approved it. - If the provider or host rejects or closes its confirmation, treat that as cancellation too.
- Silence, elapsed time, or approval for a different payload never restores authorization.
- To resume or act on a changed request, rebuild the exact preview from current state and obtain fresh explicit confirmation in a later user turn.
- After a pre-call cancellation, acknowledge that no write ran and stop this workflow unless the newest instruction explicitly replaces it with read-only work.
- If a write tool may already have started, do not claim that it was cancelled. Check status or current state, report the facts, and never retry the write automatically.
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.
- 6d ago First seen · 77 lines · 27 tokens per session scan C cd33ef6f5cea
safe-draft-and-send is a skill published in the GitHub repository ZachPulse/iblusend-plugins (0 stars, last pushed 12d ago), licensed MIT. It adds 27 tokens to every session and 906 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it C with 1 finding (nullifies safety policies). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
orbit-notion
Open Orbit briefing skill — selected by the Orbit pipeline when Notion is the user's only connected connector, or when the user explicitly scopes their daily digest to Notion. Pulls the past 24 hours of document edits, comments, mentions, and database row changes from the user's authenticated Notion connection and…
Cortex
Operate Cortex, the LifeOS memory system — the typed Knowledge Archive (People, Companies, Ideas, Research with typed related: links) plus recall of prior work sessions, ISAs, and conversations. Search, add, harvest, develop, ingest, distill, graph-navigate, recall. USE WHEN cortex, knowledge, knowledge base, search…
feishu
Work with Feishu or Lark bots, docs, sheets, bitables, approval flows, and OpenAPI/MCP setup without hardcoding credentials.
pinchtab-mcp
Use this skill when a task requires browser automation through PinchTab's MCP server connected to a remote browser instance. Covers navigation, element interaction, data extraction, form filling, multi-step flows, and session management via MCP tools.
peekaboo
Capture and automate macOS UI with the Peekaboo CLI.
mochi-remind
Handle due reminders — notify the user with natural language and mark them done.