eve Content Agent Template

A Slack-based content assistant built on eve. Writers @mention it in Slack and it drafts blog posts, LinkedIn and X posts, release notes, and newsletters in your house voice, pulling source material from Notion and publishing approved pieces back to Notion as the signed-in writer.

• Lives in Slack. Answers @mentions and DMs, replies in threads, and renders approvals as buttons.
• Writes in your voice. One editable style skill per surface (blog, LinkedIn, X, release notes, newsletter), enforced by a deterministic style-lint tool.
• Grounded in Notion. Each writer signs in to their own Notion through Vercel Connect, so drafts are created as the real person with their own permissions, with no shared secret — and page creation pauses for the writer's approval before it runs.
• Stores files in Vercel Blob. Export drafts, save images and attachments, and read them back, authenticated by the project's OIDC token.

Deploy

Deploying with the button provisions everything the agent needs and wires it up for you:

• a Slack connector (sets SLACK_CONNECTOR, with the event trigger pointed at /eve/v1/slack),
• a Notion connector (sets NOTION_CONNECTOR),
• a Vercel Blob store for the asset tools.

Once deployed, @mention the bot in your Slack workspace to start drafting.

Tech stack

Zero static keys. Authentication runs entirely on Vercel Connect (Slack and Notion) and Vercel OIDC (Vercel Blob and AI Gateway). There are no API keys or client secrets to manage in code or .env files: Notion is authorized per writer in the browser, and Blob and the model authenticate with the project's OIDC token.

Quick start with an AI coding agent

If you're working with an AI coding agent like Claude Code or Cursor, you can use this prompt to have it help you with building your agent:

I want to build a Slack agent with the eve framework, using the eve content agent template. Read the setup instructions at https://agent-resources.dev/eve-content-agent-template.md and follow them. They will cover deploying the template, building with eve, how everything works overall, and more.

What's inside

agent/
  agent.ts                  # model configuration
  instructions.md           # the agent's behavior
  channels/slack.ts         # Slack surface (Vercel Connect credentials)
  connections/notion.ts     # Notion workspace, user-scoped OAuth; page creation requires approval
  sandbox.ts                # Vercel Sandbox environment
  subagents/
    researcher/             # fresh-context web researcher (own session, web tools only)
    reviewer/               # fresh-context draft reviewer (own session); pulls its rubric via a tool
  tools/
    research.ts             # runs the researcher subagent; returns cited findings + gaps
    review.ts               # runs the reviewer subagent; returns a verdict + issues
    ask_question.ts         # ask the writer a question (Slack renders options as a menu)
    lint_against_style.ts   # deterministic banned-words check
    upload_asset.ts         # Vercel Blob: store text or binary content
    list_assets.ts          # Vercel Blob: browse stored assets
    get_asset_info.ts       # Vercel Blob: metadata without downloading
    download_asset.ts       # Vercel Blob: read a stored file back
    delete_asset.ts         # Vercel Blob: delete (requires approval)
    get_writer_preferences.ts   # load this writer's saved style preferences
    save_writer_preferences.ts  # save standing preferences (per-writer, principal-scoped)
    clear_writer_preferences.ts # clear this writer's preferences (requires approval)
  lib/
    writer-preferences.ts   # principal-scoped Blob key + reserved-prefix guard
    surfaces.generated.ts   # generated SURFACES enum (skill folders are the source of truth)
    banned-words.generated.ts # generated banned-words lists for lint_against_style
  skills/                   # one style skill per surface
    blog-style/             # + best-practices.md and format-specs.md
    linkedin-style/         # + best-practices.md and post-specs.md
    x-style/                # X (Twitter) — + best-practices.md and post-specs.md
    release-notes-style/
    newsletter-style/
shared-references/          # house-wide writing rules (source of truth), synced into each skill
  ai-phrases-to-avoid.md
  plain-english-alternatives.md
scripts/
  sync-shared.mjs           # syncs shared refs into skills; generates SURFACES, banned words + reviewer rubric (pnpm sync:shared)

Local development

Link the project you deployed (or a fresh one) and pull its environment:

vercel link
vercel env pull

Then run the development server and link a model provider with /model in the TUI:

pnpm dev

You can chat with the agent directly in the dev TUI to test the drafting, style-lint, Notion, and Blob flows. The Slack surface itself only runs against a deployment. Ship changes with:

eve deploy

Linting and formatting

This project uses Ultracite (a Biome preset) for linting and formatting:

pnpm check   # check formatting and lint rules
pnpm fix     # auto-fix what is fixable

Setting up the connectors by hand

The Deploy button provisions these for you. To set them up manually (for a project you didn't create with the button), use the Vercel CLI:

# Notion connector (note the printed UID, e.g. mcp.notion.com/notion -> NOTION_CONNECTOR)
vercel connect create mcp.notion.com --name notion

# Slack connector (note the UID, e.g. slack/<name> -> SLACK_CONNECTOR), then point its
# event trigger at the route the agent serves
vercel connect create slack --name <name> --triggers
vercel connect attach slack/<name> --triggers --trigger-path /eve/v1/slack

# Blob store, connected to the project for all environments
vercel blob create-store <name> --access public --yes

Customizing

• Voice: edit the per-surface skills in agent/skills/*/SKILL.md, and the references/banned-words.json each one lints against (run pnpm sync:shared after editing it, so lint_against_style picks up the change). Add a new surface by adding a new <surface>-style skill folder and running pnpm sync:shared — the SURFACES enum shared by lint_against_style and the reviewer is generated from the folders, so there's no list to edit.
• House-wide rules: edit the shared writing rules in shared-references/ (the source of truth), then run pnpm sync:shared. It copies them into every skill's references/ and regenerates the managed ## Shared references section in each SKILL.md (a skill's own ## References section is left alone), and regenerates agent/lib/surfaces.generated.ts, agent/lib/banned-words.generated.ts, and the reviewer's rubric module. The sync also runs automatically on pnpm dev and pnpm build. Never edit the synced copies or generated files directly.
• Behavior: edit agent/instructions.md.
• Model: edit agent/agent.ts (or run /model in the dev TUI).
• Tools: add or change tools in agent/tools/. The filename is the tool name.

The agent auto-updates as you edit these files.

Learn more

• Draft content in your voice from Slack with eve: Vercel Knowledge Base guide
• eve documentation: the framework powering this agent.
• Vercel Connect: manages the Slack and Notion credentials.
• Vercel Blob: object storage for the asset tools.

Related templates

• eve Chat Template
• eve Slack Agent
• eve Personal Agent