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 Magnus-Gille/claude-skills --skill email-triagegit clone --depth 1 https://github.com/Magnus-Gille/claude-skillsWrote 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/magnus-gille/claude-skills/email-triage)<a href="https://agentmods.dev/skills/magnus-gille/claude-skills/email-triage"><img src="https://agentmods.dev/badge/skills/magnus-gille/claude-skills/email-triage/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/magnus-gille/claude-skills/email-triage"><img src="https://agentmods.dev/badge/skills/magnus-gille/claude-skills/email-triage.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 | $0.00091 | $0.05217 |
| Opus 5 | $0.00046 | $0.02609 |
| Sonnet 5 | $0.00018 | $0.01043 |
| Haiku 4.5 | $0.00009 | $0.00522 |
Grade A, and why
email-triage 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 8d 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 — 196 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/email-triage — Reliable conversation-level email triage
Account-to-CLI map and mail/calendar search rules: ~/repos/claude-config/docs/accounts.md.
/check-email is for quick inbox glances. This skill is for trustworthy backlog triage over a long window (weeks/months), where the cost of a wrong verdict is real (chasing a closed deal, missing a live one).
The failure mode this skill exists to prevent
Naive triage judges a thread from one message's bodyPreview. That is wrong because:
- Mobile replies roll into quoted text. A reply Magnus sent from his phone often does not appear as a separate
sentitemsrow in the window — it lives inside the quoted body of a later message. Judging by inbox rows alone makes resolved threads look open. - The last inbound message is not the same as an open obligation. Martin/Saab, Koenigsegg, William — all were flagged "needs Magnus" when Magnus had already replied (or the ball was explicitly in the other party's court).
bodyPreviewtruncates before the resolution. "Thanks for this — will get back within 1–2 weeks" reads as a question in preview; the closure is past the cutoff.- A draft is not proof of an unsent obligation. Outlook leaves a lingering draft twin when you save-then-send, and old drafts from long-resolved threads persist forever. Because drafts are fetched with no date filter, a draft from months ago surfaces as "action owed" even though the offert was sent and the deal closed outside the window (Uddevalla: a March offert+acceptance was invisible to an April-onward window, so the March draft looked unsent). And a draft can carry a newer
lastModifiedDateTimethan a real sent reply in the same thread, making the sent reply look absent (Saab: an empty draft shadowed a substantive sent reply from the same morning). - The date window clips conversations. The window is a discovery filter for finding candidates, not the boundary of a conversation. A thread that looks open inside the window is frequently resolved by sent/received messages just outside it.
- A deliverable is often sent as a new thread, not a reply. Magnus answers a request by composing a fresh email with a new subject — so it has a different
conversationIdthan the request and the lingering draft. ConversationId grouping alone never connects them, so the request looks unanswered and the draft looks unsent (Photowall: the planning-meeting invite + an "Underlag inför vårt samtal" draft looked open, but the material had been sent May 31 as "AI-utbildningar Photowall — upplägg & förslag (juni)" — a separate thread to the same person). Cure: reconcile by counterparty, not just by conversationId. - The agent has no clock — it guesses dates. Without today's date stated explicitly, the subagent invents relative deadlines ("passed yesterday", "overdue") and gets them wrong (AFRY: a June-5 deadline was called overdue on June 3). Today's date must be injected, and every deadline claim must show the arithmetic.
- Editorial nudges masquerade as obligations. The agent likes to add prophylactic "worth confirming X soon" actions that no inbound message actually asked for, inflating the owed-list with invented work (Traton: "confirm final agenda" — nobody asked). Only a quotable inbound request is an obligation.
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.
- 8d ago Changed · +2 lines 21c8cd886112
- 12d ago First seen · 194 lines · 91 tokens per session scan A 130f4456ba03
email-triage is a skill published in the GitHub repository Magnus-Gille/claude-skills (5 stars, last pushed 6d ago), licensed MIT. It adds 91 tokens to every session and 5,217 once invoked, about $0.0005 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 2026-08-31.
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…
agentmail
Use your assigned AgentMail inbox to read email tasks, explicitly send or reply, and check delivery. Provided automatically by your inbox assignment.
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…
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.
feishu
Work with Feishu or Lark bots, docs, sheets, bitables, approval flows, and OpenAPI/MCP setup without hardcoding credentials.
peekaboo
Capture and automate macOS UI with the Peekaboo CLI.