
Agent Email Cli
Run the agent-email CLI from terminal workflows to create disposable inboxes, poll messages, and manage default accounts without leaking secrets.
Overview
Agent Email CLI is an agent skill for the Build phase that documents agent-email terminal commands for disposable inboxes, reads, and local profile management.
Install
npx skills add https://github.com/zaddy6/agent-email-skill --skill agent-email-cliWhat is this skill?
- Core commands: create, read, show, delete with email or default profile aliases
- Profile commands: accounts list, remove, use, config path, and about
- JSON ok/error envelope with coded errors such as ACCOUNT_NOT_FOUND
- Read supports --limit, --full, --wait, and --interval for polling automations
- Explicit rule not to surface password or token fields in agent summaries
- JSON ok/error envelope on every command
- High-signal create and read field list documented
Adoption & trust: 2.1k installs on skills.sh; 7 GitHub stars; 2/3 security scanners passed (skills.sh audits).
What problem does it solve?
Your agent needs to verify email flows but you lack a documented, JSON-safe CLI recipe for create-read-show-delete automation.
Who is it for?
Builders automating auth and notification tests with terminal agents and jq-friendly output.
Skip if: Production transactional email delivery, marketing inboxes, or teams that forbid local CLI tools with network access.
When should I use this skill?
Operate agent-email CLI for mailbox creation and message reads in terminal agent workflows.
What do I get? / Deliverables
Your agent runs standardized agent-email commands and parses JSON envelopes to complete inbox tests without exposing credentials in chat.
- Parsed message ids and subjects from JSON
- Disposable inbox addresses for test runs
Recommended Skills
Journey fit
Build agent-tooling is the primary shelf because the skill documents how coding agents invoke mailbox automation during integration and E2E flows. Agent-tooling fits disposable inbox creation, JSON envelopes, and jq-friendly automation snippets aimed at agent-driven test and signup flows.
How it compares
Terminal integration skill for disposable test mail—not an MCP server or SendGrid-style ESP integration.
Common Questions / FAQ
Who is agent-email-cli for?
Solo builders and agent users who test signup and magic-link flows via the agent-email CLI in the terminal.
When should I use agent-email-cli?
During Build when wiring agent-tooling or integrations that must create inboxes, poll for mail, and delete test messages.
Is agent-email-cli safe to install?
The skill handles real mailbox credentials locally—review the Security Audits panel on this Prism page and never paste secrets into agent logs.
SKILL.md
READMESKILL.md - Agent Email Cli
interface: display_name: "Agent Email CLI" short_description: "Operate agent-email CLI for mailbox creation and message reads." default_prompt: "Use this skill to run agent-email commands for creating disposable inboxes, polling for incoming mail, retrieving message details, and managing local account profiles in terminal agent workflows." # Command Reference ## Core commands ```bash agent-email create agent-email read <email|default> [--limit <n>] [--full] [--wait <s>] [--interval <s>] agent-email show <email|default> <messageId> agent-email delete <email|default> <messageId> ``` ## Profile commands ```bash agent-email accounts list agent-email accounts remove <email|default> agent-email use <email|default> agent-email config path agent-email about ``` ## Output envelope All commands return JSON by default: ```json { "ok": true, "command": "create", "data": {} } ``` or on failure: ```json { "ok": false, "command": "read", "error": { "code": "ACCOUNT_NOT_FOUND", "message": "...", "hint": "..." } } ``` ## High-signal fields - `create`: `data.email`, `data.accountId`, `data.activeEmail` - `read`: `data.messages[]` with `id`, `from.address`, `subject`, `createdAt`, `seen` - `show`: `data.message`, optional `data.source` Do not extract or print secret fields (`password`, `token`) in summaries. ## Common automation snippets Read active inbox and extract first message id: ```bash agent-email read default | jq -r '.data.messages[0].id' ``` Fetch full first message: ```bash MSG_ID=$(agent-email read default | jq -r '.data.messages[0].id') agent-email show default "$MSG_ID" ``` --- name: agent-email-cli description: Operate the agent-email CLI to create disposable inboxes, poll for new mail, retrieve full message details, and manage local mailbox profiles. Use when the user needs terminal-based email inbox access for LLM or agent automation workflows. --- # Agent Email CLI ## Overview Use this skill to operate the `agent-email` command safely and predictably for agent workflows that need inbox access. Prefer JSON-native command output and return key fields (`email`, `messageId`, `subject`, `createdAt`, `from.address`) in your summaries. ## Workflow 1. Verify CLI availability. ```bash command -v agent-email agent-email --help ``` If missing, install: ```bash npm install -g @zaddy6/agentemail # or bun install -g @zaddy6/agentemail ``` 2. Create a mailbox account. ```bash agent-email create ``` Record these fields from JSON output: - `data.email` - `data.accountId` - `data.activeEmail` Do not record, repeat, or print secret values such as mailbox passwords or tokens. 3. Read latest messages. ```bash agent-email read <email|default> ``` For inbox waiting/polling: ```bash agent-email read <email|default> --wait 30 --interval 2 ``` For full message payloads: ```bash agent-email read <email|default> --full ``` 4. Retrieve one message in detail. ```bash agent-email show <email|default> <messageId> ``` Use `show` when you need body/source details for verification links, codes, or full content extraction. 5. Manage mailbox profiles. ```bash agent-email accounts list agent-email use <email|default> agent-email accounts remove <email> ``` Avoid commands that require entering secrets on the command line in agent logs. 6. Delete processed/irrelevant message when requested. ```bash agent-email delete <email|default> <messageId> ``` ## Operational Guidance - Keep command output machine-readable; avoid forcing human output unless requested. - Prefer `default` alias when user does not specify an email. - Never echo, store, or summarize secret values (`password`, `token`) from command output. - If command fails, surface the JSON error `code` and `hint` fields directly. - For auth failures (`AUTH_REQUIRED`/401), rerun command once and request user intervention if credentials must be re-established. - For rate limits (`RATE_LIMITED`/429), retry after short delay. ## Trouble