brainstack

engineer-brain — onboarding

Who this is for: anyone using engineer-brain sync for daily standups.
Time: ~15 minutes (one-time setup).
You need: an AI assistant with Brainstack installed, git repos in your workspace, GitHub CLI, and a Jira signal (see below).

What gets configured

Signal	Required?	Setup
Git history	Yes	Clone team repos under your workspace
GitHub PRs/reviews	Yes	`gh auth login`
Jira tickets	Yes	Atlassian MCP (Cursor) or `jira.sh` (other platforms)
Google Calendar	Optional	gcal MCP — demos, workshops, meetups
`BRAIN.md`	Yes (identity)	Fill name/role/team after `install.sh`

sync always pulls Jira. Do not run standup prep without a working Jira signal.

Step 1 — Install engineer-brain

git clone https://github.com/Hrithik-Gavankar/brainstack.git
cd brainstack
bash install.sh cursor ~/path/to/your-workspace   # or claude-code, vscode-copilot, etc.

Verify (Cursor example):

ls ~/.cursor/skills/engineer-brain/SKILL.md
ls ~/.cursor/skills/engineer-brain/ONBOARDING.md
ls ~/.cursor/rules/engineer-brain.mdc

Open BRAIN.md in your platform skill directory (or .engineer-brain/BRAIN.md) and fill Identity + Current Sprint Context.

Step 2 — GitHub CLI

gh auth login
gh auth status

Used by scan.sh for authored PRs, reviews, and releases.

Step 3 — Jira signal (required)

Cursor — Atlassian MCP (recommended)

Jira is not configured in ~/.cursor/mcp.json. Use the official Cursor marketplace plugin (OAuth — no API token in shell).

3.1 Install the plugin

Open Cursor.
Go to Settings → Plugins (or open cursor.com/marketplace/atlassian).
Find Atlassian (by Atlassian) and click Install / Enable.
Reload Cursor if prompted.

The plugin registers MCP namespace plugin-atlassian-atlassian and points at Atlassian’s hosted server (https://mcp.atlassian.com/v1/mcp/authv2). You do not add a manual mcpServers block for Jira.

3.2 Authorize (first use)

Start a Cursor chat in your workspace.
Ask: “Who am I on Atlassian?” or run /engineer-brain sync.
When the OAuth browser window opens, sign in with your company Atlassian account (e.g. you@company.com on your-org.atlassian.net).
Approve the requested Jira/Confluence scopes.

First-time site install may require a user who has access to the Jira projects you use; after that, other crew members can authorize individually.

Example (Red Hat): you@redhat.com on redhat.atlassian.net, project keys like YOU_JIRA_TICKET_HERE.

3.3 Verify

In chat, the agent should succeed on:

atlassianUserInfo — returns your name and email
getAccessibleAtlassianResources — returns your site cloud id (e.g. your-org.atlassian.net)
searchJiraIssuesUsingJql — returns your assigned issues (requires cloudId from the previous call)

If the agent reports no Jira namespace or needsAuth:

Symptom	Fix
Plugin not listed in Settings → Plugins	Re-install from marketplace; restart Cursor
OAuth never appeared	Retry a Jira tool call; check pop-up blocker
`401` / expired session	Re-authenticate via plugin settings or repeat OAuth flow
Org blocks Rovo MCP	Ask your Atlassian admin — control settings

3.4 What `sync` queries (automatic)

When you run /engineer-brain sync, the agent must:

Resolve cloudId via getAccessibleAtlassianResources
Run JQL for tickets updated in the standup window (yesterday, or Friday-only on Mondays)
Run JQL for your active assignments (statusCategory = "In Progress")
Optionally check open sprint (sprint in openSprints()) — many teams do not sprint-tag every ticket; the in-progress query is the fallback

Sample JQL (replace CLOUD_ID with the id from getAccessibleAtlassianResources):

# Tuesday–Friday: tickets you touched yesterday
assignee = currentUser() AND updated >= startOfDay(-1) ORDER BY updated DESC

# Monday standup: Friday only
assignee = currentUser() AND updated >= startOfDay(-3) AND updated < startOfDay(-1) ORDER BY updated DESC

# Carry-forward for "today" section
assignee = currentUser() AND statusCategory = "In Progress" ORDER BY updated DESC

Tickets surface in standup as PROJ-12345 (your project key) with impact language, not raw JQL dumps.

Other platforms — `jira.sh` CLI

Claude Code, Copilot, Windsurf, Aider, and Continue.dev do not have the Atlassian Cursor plugin. For sync on those platforms, configure the CLI fallback (Step 5) and ensure install.sh copied jira.sh into your .engineer-brain/scripts/ (or platform skill scripts/).

Step 4 — Optional: Google Calendar

For hackathons, demos, workshops, and meetups that never appear in git:

See gcal MCP setup — one-time OAuth via gcal.sh authorize.

Step 5 — Optional: `jira.sh` CLI fallback

For terminals, non-Cursor platforms, and CI — not a substitute for Atlassian MCP in Cursor when the plugin is connected.

export JIRA_URL="https://your-org.atlassian.net"
export JIRA_EMAIL="you@company.com"
export JIRA_API_TOKEN="..."   # https://id.atlassian.com/manage-profile/security/api-tokens
bash .engineer-brain/scripts/jira.sh active

Add the exports to ~/.zshrc if you want shell scripts to work. Cursor sync prefers Atlassian MCP when the plugin is connected.

Daily use

/engineer-brain sync

Or: “Help me with sync up notes for today.”

Expected sources in every run: git + gh + Jira + BRAIN.md + gcal (if configured).

core/COMMANDS.md — full command reference
team-brain-onboarding.md — crew shared memory (separate from personal standup)

This site is open source. Improve this page.