Jira (jira)

The jira connector indexes Jira issues by project, plus the user directory. A search hit covers the whole issue conversation — summary, description, and comments — so you find an issue by what was discussed, not just its title.

How MFS sees it

jira://acme/
├── projects/
│   ├── ENG/issues.jsonl     record_collection
│   └── OPS/issues.jsonl
└── users.jsonl              record_collection

The jira.issues preset embeds summary, description, and comment bodies, with status/priority/assignee as metadata and key as the locator — so issues are searchable without [[objects]] config, though you can override the fields.

Credentials

Three flavours, by deployment:

• Atlassian Cloud (most common): URL https://acme.atlassian.net, username = your account email, and an API token from https://id.atlassian.com/manage-profile/security/api-tokens → Create API token.
• Server / Data Center (self-hosted): URL of your instance, username left empty, and a Personal Access Token from your Jira profile.
• Older Server (no PAT): username + password basic auth (discouraged).

An API token inherits the issuing user's permissions — restricted projects look empty if that user can't see them.

Configuration

url = "https://acme.atlassian.net"
cloud = true
username = "alice@acme.com"
api_token = "env:JIRA_API_TOKEN"
projects = ["ENG", "OPS"]      # empty = all visible projects
max_read_rows = 50000

Sync and freshness

The connector uses the issue updated timestamp as its cursor; Cloud retrieval pages through enhanced_jql. Deletions are caught by full_scan.

Search and browse

mfs add jira://acme --config ./jira.toml

mfs search "SSO regression" jira://acme/projects/ENG/issues.jsonl
mfs cat jira://acme/projects/ENG/issues.jsonl --locator '{"key":"ENG-1234"}'

Pitfalls

• Without projects, the connector enumerates every visible project.
• The flattened record uses key as the issue-key field.
• Token permissions are the user's permissions; restricted projects appear empty.