Email Summary & Evidence-linked Action Items
Read authorized mailboxes without changing their state. Produce Chinese briefs with source-linked decisions and action items.
Free
Works with the AI Agents you already use
Claude Code
Cursor
Codex CLI
Muse
Dots+22 moreClaude CodeCursorCodex CLIMuseDots
OpenClaw
Manus
Grok
Grok Bot+18 agentsEmail Summary & Evidence-linked Action Items
Example session with this skill installed
Use email-summary-action-items with the package's two fictional example emails. Show the expected Chinese brief, identify the action and the commitment, and retain message evidence references. Label it as a fictional offline example. Do not connect to a real mailbox or call a model for this demonstration.
- Read your context and instructions
- Compiled the email summary
Fictional offline example
The report below is an illustrative expected output for the two bundled fictional emails, with an unsupported source detail removed. It is not a live-mailbox session or a model-generated result from this demonstration. The example URLs use example.com and do not open a real mailbox.
2026-09-07 邮件总结与待办
今日速览
- 项目评审已经通过；需要在周三前发送修改后的文档。[work/example-001]
需要处理
- 修改并发送项目文档。负责人：我；截止时间：周三，具体日期未说明。[work/example-001] [work/example-002]
已确认决策与承诺
- 我方已经承诺在周三前发送修改后的文档。[work/example-002]
等待他人
无。
重要信息
- 项目评审已经通过。[work/example-001]
低优先级
无。
来源与覆盖限制
- 覆盖虚构邮箱
work中thread-project-review会话的两封示例邮件。 - 查看虚构来源链接。
- 示例没有给出“周三”的具体日期，因此未推断日期。
Checks performed
- Normalized messages: valid; 2 messages in 1 thread.
- Writer dry run: 1,201 prompt characters; within the configured input limit.
- Summary structure and evidence IDs: VALID.
- These checks do not establish model accuracy or a live provider integration.
Connects securely to your tools. The creator never sees your data.
About this skill
What it does
Turn authorized email messages into a Chinese Markdown brief with decisions, commitments, requested actions, owners and deadlines when stated, and references to the supporting messages.
- Collect from configured IMAP folders using read-only selection and BODY.PEEK[], or use an existing authorized connector.
- Group collected messages by account and thread; report collection errors and known missing thread context.
- Validate message data before model use and check report sections and evidence IDs afterward.
- Keep source links when the input provides them. The bundled IMAP collector does not generate webmail links.
Requirements and data flow
Requires Python 3.10+, an agent with local terminal and task-file access, authorized mailbox access, and an OpenAI-compatible Chat Completions model endpoint. Selected email content is sent to the model endpoint you authorize. Your model provider may charge usage fees. The current writer outputs Chinese Markdown.
The bundled collector uses IMAP over TLS with username/password login; it does not implement OAuth. Providers requiring OAuth need a separate authorized connector. Credentials stay in your local environment or existing credential mechanism.
Agensi use
The package includes four independent Python scripts, message contracts, report templates and fictional examples. For MCP delivery, the agent reads supporting file text from the response and executes scripts through standard input without saving Skill source files. User configuration, mailbox data, summaries and validation reports are separate task files.
Scope and limitations
Default collection limits are 200 messages per folder and 6,000 body characters per message. A time-window collection may omit thread history. Attachment names are listed; attachment content is not extracted. Evidence-ID validation does not prove that a summary is semantically correct, so review decisions and deadlines against the source text.
Local offline checks and four upstream unit tests passed. IMAP and model calls in the unit tests are mocked. Real mailbox login, a live model request and the Agensi buyer flow have not been verified.
Source and license
Free listing. Public source remains available under MIT at https://github.com/xiangri678/email-summary-action-items-skill. The original LICENSE is included. This distribution adds Agensi execution guidance and corrects one unsupported detail in a fictional example; the four Python scripts are unchanged.
How to use
Skills on Agensi run inside your AI agent. Copy the prompt, paste it into your agent, and it connects to Agensi and installs this skill for you. If a payment is needed, your agent shows it to you first.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scanned
Verified clean 3 days ago
- Free to download with an account
Requires
Python