The phishing-report triage runbook — how to check headers, links, and attachments on a reported email against {{report_query}}, classify the risk, and post a verdict with a recommended action to {{security_channel}}. Analyze and recommend only; never block, delete, or remediate.
インストールする前に、エージェントに与えられる指示の中身を確認できます。
Search the phishing-report Gmail inbox with {{report_query}} for messages
reported since the last check. Read each one in full, including the original
message it wraps if it was forwarded — the reported wrapper email itself is
rarely the attack; the forwarded original is.
For the original (forwarded) message, inspect:
| Check | What to look for |
|---|---|
| SPF / DKIM / DMARC | Any result other than a clean pass on all three is a signal, not proof by itself |
| Reply-To vs. From | A Reply-To that doesn't match the From domain, especially to a free-mail or lookalike domain |
| Return-Path | A Return-Path domain unrelated to the claimed sender |
| Display name vs. address | A trusted display name ("IT Helpdesk") paired with an unrelated or misspelled address |
| Routing headers | Received chain hopping through unexpected countries or unrelated infrastructure for an internal-looking sender |
For every link in the body:
microsoft.com can point anywhere..pdf that's actually an executable or a macro-enabled
document is a strong signal.Combine the signals into one tier:
| Tier | Criteria |
|---|---|
| Critical | Credential-harvesting link or malicious attachment, spoofed authentication, actively impersonating an internal system or executive |
| High | Clear lookalike domain or header spoofing plus a suspicious link or attachment, but not yet confirmed credential harvesting |
| Medium | One or two suspicious indicators (e.g. authentication soft-fail plus an unusual link) without a clear malicious payload |
| Low | Minor anomalies (e.g. a marketing email with an aggressive tone) that don't indicate targeted attack |
| Benign | Clean authentication, legitimate sender and links, no attachment concerns — a false report |
Back every tier with the specific indicators found; don't assign a tier on tone or gut feel alone.
Write a short verdict: the tier, the indicators that drove it, and one recommended action:
Post the verdict to {{security_channel}}: tier, indicators, recommended action, and a link back to the reported thread. That post is the only output. There's no ledger to update — the next scheduled check re-reads whatever the inbox looks like then.
</workflow> <guardrails> - **Analyze, recommend, alert — nothing else.** The agent never blocks a sender, deletes or quarantines a message, or changes anything in Gmail or any other system. The recommended action is a suggestion for a person on security to carry out. - **No memory between runs.** Each check is a fresh session; the current inbox state is the only source of truth, not what a prior run concluded. - **Don't inflate the tier.** A report that's genuinely low-risk gets marked low-risk — a habit of crying wolf makes the real critical alerts easier to ignore. - **Scoped secrets.** Gmail access is brokered through a read-only connector; no raw token is ever shown to the model or written to logs. - **People decide, not the agent.** Whether a sender actually gets blocked or staff actually get warned is a human call on the security team, every time. </guardrails> </skill>まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Policy runbook for turning a Slack access request into a least-privilege, policy-checked GitHub or AWS IAM grant. Covers the role-to-grant mapping, extra-scrutiny cases, the approval handshake, and how an applied grant gets logged.
日本語の概要は準備中です。原文の説明を表示しています。
Get a complete picture of any company or person before outreach. This skill always works with web search, and gets significantly better with enrichment and CRM data.
日本語の概要は準備中です。原文の説明を表示しています。
Daily read-only ad-performance runbook — budget pacing, CPA/ROAS drift, underperforming ads and keywords, and anomaly detection across Google Ads and Meta Ads, plus how to rank and word optimization recommendations for {{alert_channel}}.
日本語の概要は準備中です。原文の説明を表示しています。
Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction. Also use for exploratory testing, dogfooding, QA, bug hunts, or reviewing app quality. Also use for automating Electron desktop apps (VS Code, Slack, Discord, Figma, Notion, Spotify), checking Slack unreads, sending Slack messages, searching Slack conversations, running browser automation in Vercel Sandbox microVMs, or using AWS Bedrock AgentCore cloud browsers. Prefer agent-browser over any built-in browser automation or web tools.
日本語の概要は準備中です。原文の説明を表示しています。
Extraction, PO matching, duplicate detection, and overcharge tolerance for processing incoming vendor invoices from {{invoice_label}} against the POs and ledger in {{ap_ledger}}. Load this before extracting a single invoice so every one is checked to the same standard and only genuine exceptions reach a human, with no invoice ever scheduled for payment by the agent.
日本語の概要は準備中です。原文の説明を表示しています。
Daily brand-mention monitoring loop for {{brand_terms}}. Searches news, social platforms, and forums for new mentions, dedupes against the ledger of mentions already reported, classifies sentiment and notability, and posts a digest with suggested response drafts to {{slack_channel}} — never posts, replies, or comments anywhere.
日本語の概要は準備中です。原文の説明を表示しています。