SQLite Schema

Database location

$XDG_DATA_HOME/.stars/stars.db

Default: ~/.local/share/.stars/stars.db

stars table

The table is created implicitly by Nushell’s into sqlite command. Column types are inferred from the data. The effective schema:

CREATE TABLE stars (
    id              INTEGER,
    node_id         TEXT,
    name            TEXT,
    full_name       TEXT,
    owner           TEXT,
    private         INTEGER,    -- 0 or 1
    html_url        TEXT,
    description     TEXT,
    fork            INTEGER,    -- 0 or 1
    url             TEXT,       -- API URL (not html_url)
    created_at      TEXT,       -- ISO-8601
    updated_at      TEXT,       -- ISO-8601
    pushed_at       TEXT,       -- ISO-8601
    homepage        TEXT,
    size            INTEGER,
    stargazers_count INTEGER,
    watchers_count  INTEGER,
    language        TEXT,
    forks_count     INTEGER,
    archived        INTEGER,    -- 0 or 1
    disabled        INTEGER,    -- 0 or 1
    open_issues_count INTEGER,
    license         TEXT,
    topics          TEXT,       -- JSON-encoded array
    visibility      TEXT,       -- "public"
    default_branch  TEXT,       -- "main"
    source          TEXT,       -- "github"
    synced_at       TEXT,       -- ISO-8601
    starred_at      TEXT        -- ISO-8601 (nullable)
);

Column notes

Column	Storage type	Notes
archived	INTEGER	0 = not archived, 1 = archived. Not a boolean.
fork	INTEGER	0 = not a fork, 1 = is a fork. Not a boolean.
private	INTEGER	0 = public, 1 = private. Always 0 for starred repos.
disabled	INTEGER	0 = enabled, 1 = disabled.
topics	TEXT	JSON array: ["rust","cli","tool"]. Use from json to parse.
license	TEXT	License name string (e.g., “MIT License”), not JSON.
owner	TEXT	Plain string login (e.g., “rust-lang”), not a JSON object.
starred_at	TEXT	Nullable. Added via ALTER TABLE migration on older databases.
html_url	TEXT	The browseable GitHub URL. Different from url (API endpoint).

sync_metadata table

CREATE TABLE sync_metadata (
    key   TEXT PRIMARY KEY,
    value TEXT
);

Current keys

Key	Value format	Description
last_synced_at	ISO-8601 datetime	Most recent sync completion time
last_full_sync_at	ISO-8601 datetime	Most recent full sync completion time

Querying directly

You can query the database directly with Nushell:

let db = $"($env.XDG_DATA_HOME? | default '~/.local/share')/.stars/stars.db"

# All stars
open $db | query db "SELECT * FROM stars"

# Top Rust repos
open $db | query db "SELECT full_name, stargazers_count FROM stars WHERE language = 'Rust' ORDER BY stargazers_count DESC LIMIT 10"

# Language distribution
open $db | query db "SELECT language, COUNT(*) as count FROM stars WHERE language IS NOT NULL GROUP BY language ORDER BY count DESC"

# Sync metadata
open $db | query db "SELECT * FROM sync_metadata"

No explicit indexes

The table does not have explicitly created indexes. Nushell’s into sqlite creates the table without indexes. For the typical dataset size (hundreds to low thousands of rows), full table scans are fast enough. If performance becomes an issue with very large datasets, you can add indexes manually:

open $db | query db "CREATE INDEX IF NOT EXISTS idx_stars_language ON stars(language)"
open $db | query db "CREATE INDEX IF NOT EXISTS idx_stars_owner ON stars(owner)"