Cursor Skills

A collection of AI agent skills for Cursor that integrate with Slack and Jira to automate daily workflows.

Available Skills

Skill	Description
Daily Digest	- Summarizes your Slack channels, DMs, and Jira mentions into a browsable daily report. - Track actions for you and your team members, which you can mark as done. - Track Features in Jira which you want to watch for progress & get summary updates that span Jira and Slack.

Daily Digest: what you get

After you run a digest (e.g. "Give me a daily digest for yesterday" in Agent chat):

• digests/YYYY-MM-DD.json — Structured digest for that day (under .cursor/skills/daily-digest/digests/)
• HTML viewer — Browse all digests, actions, and feature tracking in the browser (viewer/)
• Markdown export — YYYY-MM-DD.md for Obsidian or other notes apps (default: markdown/; optional vault path via obsidian_digest_path in user_config.json)
• Jira integration — Feature watchlist + @mention tracking in the viewer (active_features.json, jira_mentions.json; see SKILL.md)

Full workflow, edge cases (person-only search, safe merge when MCP fails), and step-by-step instructions are in Daily Digest SKILL.md. A short skill overview lives in .cursor/skills/daily-digest/README.md.

Architecture

The diagram below shows how information flows from Slack and Jira through MCP servers, into the Daily Digest Skill, and out to the HTML viewer and Obsidian:

Prerequisites

• Cursor IDE with Agent mode enabled
• A Slack workspace you can generate tokens for
• A Jira Cloud instance with API token access
• npx and uvx on your PATH (Node.js and Python/uv) — used by MCP servers
• python3 — used to export each digest to Markdown (export_digest_to_md.py) at the end of a digest run (Step 7 in the skill)

Quick Start

1. Clone the repo

git clone <repo-url> cursor-skills
cd cursor-skills

Open this folder as the workspace root in Cursor (the root that contains user_config.json).

2. Configure MCP servers

The skills talk to Slack and Jira through MCP (Model Context Protocol) servers running locally. Copy the example config and fill in your credentials:

cp .cursor/mcp.example.json .cursor/mcp.json

Edit .cursor/mcp.json with your values:

{
  "mcpServers": {
    "user-atlassian": {
      "type": "stdio",
      "command": "uvx",
      "args": ["mcp-atlassian"],
      "env": {
        "JIRA_URL": "https://your-company.atlassian.net",
        "JIRA_USERNAME": "you@company.com",
        "JIRA_API_TOKEN": "<your-jira-api-token>",
        "JIRA_SSL_VERIFY": "true"
      }
    },
    "user-slack": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "slack-mcp-server"],
      "env": {
        "SLACK_MCP_XOXC_TOKEN": "<your-xoxc-token>",
        "SLACK_MCP_XOXD_TOKEN": "<your-xoxd-token>",
        "SLACK_MCP_CUSTOM_TLS": "1"
      }
    }
  }
}

Server names: The example uses user-slack and user-atlassian. Your Cursor MCP config may use different keys (e.g. user-user-slack). The agent must call tools on the servers you actually configured—if the skill references other names, align either mcp.json or the skill’s workflow. If tools fail with “server not found,” check names in Cursor Settings → MCP.

Getting your tokens:

• Jira API token — generate one at id.atlassian.com/manage-profile/security/api-tokens
• Slack tokens (xoxc / xoxd) — these are session tokens extracted from the Slack web app. Open Slack in your browser, open DevTools, and find them in cookies/local storage. They rotate, so you may need to refresh them periodically.

3. Set up your identity

Copy the example user config and fill in your details:

cp user_config.example.json user_config.json

Edit user_config.json. The template matches user_config.example.json:

{
  "jira_display_name": "Your Full Name",
  "jira_username": "your-jira-username",
  "jira_email": "you@company.com",
  "jira_account_id": "712020:xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx",
  "slack_username": "your-slack-username",
  "slack_domain": "your-company.enterprise.slack.com",
  "slack_dm_domain": "your-company-internal.slack.com",
  "jira_base_url": "https://your-company.atlassian.net",
  "gh_default_repo": "your-org/your-repo",
  "obsidian_digest_path": ""
}

Key	Purpose
slack_username	Your Slack username (no @) — used for self-DM and “my actions”
slack_domain	Host for channel Slack archive links
slack_dm_domain	Host for DM thread links (often different from slack_domain in large orgs). Omit only if DMs use the same host as channels
jira_display_name / jira_username	Shown in summaries and fallback Jira search
jira_email	Optional; helps resolve jira_account_id if missing
jira_account_id	Recommended — Atlassian account ID for reliable Jira @mention search in comments (comment ~ "712020:…")
jira_base_url	Your Jira Cloud base URL
gh_default_repo	Optional default GitHub repo for tooling
obsidian_digest_path	Optional absolute path to a folder in your Obsidian vault. If empty, Markdown exports go to .cursor/skills/daily-digest/markdown/

This file is shared across all skills and is gitignored.

4. Configure the Daily Digest (optional but recommended)

Set up which Slack channels to monitor:

cp .cursor/skills/daily-digest/slack_channels_config.example.json \
   .cursor/skills/daily-digest/slack_channels_config.json

Edit slack_channels_config.json to list your channels, usergroup mentions, and whether to include DMs:

{
  "channels": ["#your-team-channel", "#your-working-group"],
  "mention_groups": ["@your-usergroup"],
  "include_dms": true
}

Optionally, set up the Jira feature watchlist to track feature progress alongside your digest:

cp .cursor/skills/daily-digest/feature_watchlist.example.json \
   .cursor/skills/daily-digest/feature_watchlist.json

Add your Jira feature keys to feature_watchlist.json.

5. Open the project in Cursor and go

Open the cursor-skills folder in Cursor. The skills are automatically discovered. Try:

"Give me a daily digest for yesterday"

Viewing digests

HTML viewer

After generating a digest, serve the built-in HTML viewer:

npx serve .cursor/skills/daily-digest

Then open http://localhost:3000/viewer/ in your browser (port may differ). The viewer shows all generated digests in a browsable timeline with action items, thread links, and feature tracking.

Obsidian / Markdown

Each successful digest run should produce a Markdown file for that date via export_digest_to_md.py (see Step 7 in SKILL.md):

• Default output: .cursor/skills/daily-digest/markdown/YYYY-MM-DD.md
• If obsidian_digest_path in user_config.json is set to an absolute path inside your vault, the .md file is written there instead (you can also symlink markdown/ into a vault)

Requires python3 on your PATH. Backfill all digests once:

python3 .cursor/skills/daily-digest/export_digest_to_md.py --all

Project structure

cursor-skills/
├── README.md
├── user_config.json              # Your identity (gitignored, create from .example)
├── user_config.example.json      # Template
└── .cursor/
    ├── mcp.json                  # MCP server config (gitignored, create from .example)
    ├── mcp.example.json          # Template
    └── skills/
        └── daily-digest/
            ├── README.md         # Short pointer + quick links
            ├── SKILL.md          # Full workflow (steps, MCP, safe merge, person search)
            ├── export_digest_to_md.py
            ├── slack_channels_config.json
            ├── feature_watchlist.json
            ├── active_features.json    # Auto-generated (Jira feature tracker)
            ├── jira_mentions.json      # Auto-generated (merged mention history)
            ├── markdown/               # Default Obsidian-ready .md exports
            ├── digests/                # Auto-generated daily JSON
            └── viewer/                 # Self-contained HTML digest viewer

Troubleshooting

MCP servers not connecting — Restart Cursor after editing .cursor/mcp.json. Check that npx and uvx are on your PATH by running them in your terminal. Confirm server names in mcp.json match what Cursor exposes (see Server names under Configure MCP servers).

Slack tokens expired — The xoxc/xoxd tokens are session-based and expire. Re-extract them from the Slack web app when you see auth errors.

"Could not load user_config.json" — Make sure you copied user_config.example.json to user_config.json in the workspace root (not inside the skill directory).

Empty digest — Verify your slack_channels_config.json lists channels you actually have access to, and that the date you requested has activity. If Slack MCP was not authenticated, you may get an empty or partial digest; see Safe write rules in SKILL.md.

Markdown file missing — Ensure python3 works (python3 --version). Re-run: python3 .cursor/skills/daily-digest/export_digest_to_md.py .cursor/skills/daily-digest/digests/YYYY-MM-DD.json from the repo root.