@plurnk/plurnk-execs-sqlite

SQLite runtime executor for plurnk-service's exec scheme. Runs the sqlite runtime tag in-process via Node's builtin node:sqlite — no subprocess.

A @plurnk/plurnk-execs-* sibling built on the plurnk-execs framework.

Runtime tag

| Tag | Glyph | Engine | |---|---|---| | sqlite | 🗃 | node:sqlite (Node 25 builtin) |

Database target

The EXEC target slot is the database file; with no target it defaults to an ephemeral in-memory db:

<<EXEC[sqlite]:SELECT * FROM users:EXEC            → :memory: (fresh per run)
<<EXEC[sqlite](./app.db):SELECT * FROM users:EXEC  → ./app.db (persistent)

:memory: is ephemeral — state does not persist across EXECs. Pass a file path for persistence.

Output

Writes to the results channel as application/json, ready for the jsonpath body-matcher (plurnk-mimetypes' JSON handler):

• Row-returning statements (SELECT, RETURNING, PRAGMA) → an array of row objects.
• Mutations (INSERT/UPDATE/DELETE/DDL) → { changes, lastInsertRowid }.

The query/mutation split is decided by the prepared statement's columns(), never by parsing the SQL. One statement per EXEC. Errors emit a TelemetryEvent (source: "exec:sqlite"): sqlite_open_failed, sqlite_error.

Availability & proposal gating

probe() always reports available (node:sqlite is a builtin). effect(target) marks :memory: (and no target) as pure (auto-run) and a file-backed db as host (propose) — classified by the target only, never by inspecting the SQL.

Tests

test:lint, test:unit.