DeltaScope

DeltaScope is an offline-first SQL audit engine for MySQL, TiDB, and PostgreSQL. The main product surfaces are deltascope, deltascope-server, and deltascope-mcp; PostgreSQL offline support is converged on the main archives for the supported macOS and Linux platforms instead of living behind a separate PG-only CLI entrypoint. As of v0.28.0, DeltaScope exposes PostgreSQL referenced-object facts (referenced_schema, referenced_table, referenced_columns) in FK forbid finding metadata — an additive metadata widening across all four transport surfaces. It gives DBAs, application engineers, CI pipelines, and AI agents one consistent way to review DDL and DML before they reach a database.

Install

For macOS, prefer Homebrew. The repository installer script remains available as the generic portable installer for environments where Homebrew is not the right fit.

macOS (recommended):

brew tap Fanduzi/deltascope
brew install --cask deltascope

Generic installer:

curl -fsSL https://raw.githubusercontent.com/Fanduzi/DeltaScope/main/install.sh | sh

Pin a specific release:

curl -fsSL https://raw.githubusercontent.com/Fanduzi/DeltaScope/v0.28.0/install.sh | \
  DELTASCOPE_VERSION=v0.28.0 sh

Referenced-Object Metadata Surface Pack (v0.28.0)

v0.28.0 exposes PostgreSQL referenced-object facts (referenced_schema, referenced_table, referenced_columns) as additive finding metadata on the FK forbid rule, across CLI, HTTP, MCP, and pkg/deltascope. It does not add new rules, new CLI flags, or new public API contracts, and it is not schema-aware FK policy support.

• Finding metadata widening: the ddl.table.foreign_key.forbid finding now includes referenced_schema (e.g., "public"), referenced_table (e.g., "users"), and referenced_columns (e.g., ["id"]) when the underlying constraint carries those facts. referenced_table is never concatenated into "public.users".
• Parser/extractor semantics are unchanged from v0.27.0. The shared semantic contract (spec.Constraint) already had ReferencedSchema, ReferencedTable, and ReferencedColumns; v0.28.0 widens the outward finding metadata to expose them.
• No new rule IDs, no new CLI flags, no schema-aware FK policy decisions.

Previous milestone: v0.27.0 preserved schema-qualified PostgreSQL referenced-object facts in the shared contract. See the v0.27.0 release notes for details.

Schema-Qualified Reference Semantics Pack (v0.27.0)

v0.27.0 preserves schema-qualified PostgreSQL referenced-object facts in the shared contract. It does not add new rules, new CLI flags, or new public API contracts, and it is not full PostgreSQL foreign key support.

• ReferencedSchema is an additive field on spec.Constraint that preserves the schema portion of schema-qualified REFERENCES (e.g., REFERENCES public.users(id) → ReferencedSchema = "public", ReferencedTable = "users").
• PostgreSQL extractor preserves these facts for both named FOREIGN KEY and inline REFERENCES forms.
• Corpus and service-level tests lock the semantic contract with precise assertions.
• CLI, HTTP, MCP, and pkg/deltascope current public finding metadata remains unchanged — the shared semantic contract is richer underneath.

Previous milestone: v0.26.0 tightened the PostgreSQL CREATE TABLE unsupported boundary contract. See the v0.26.0 release notes for details.

PostgreSQL CREATE TABLE Unsupported Boundary Pack (v0.26.0)

v0.26.0 tightens the PostgreSQL CREATE TABLE unsupported boundary contract. It does not add new rules, new CLI flags, or new public API contracts, and it is not full PostgreSQL CREATE TABLE support.

• Identity columns (GENERATED ... AS IDENTITY) → explicit unsupported (generated_as_identity).
• Generated stored columns (GENERATED ALWAYS AS ... STORED) → explicit unsupported (generated_column).
• Exclusion constraints (EXCLUDE USING) → explicit unsupported (exclusion_constraint).
• Partitioned tables (PARTITION BY) → explicit unsupported (partitioning).
• PostgreSQL corpus cases lock these four boundaries with precise expected-outcome assertions.
• Surface parity tests across CLI, HTTP, MCP, and pkg/deltascope verify each boundary is exposed through the correct unsupported contract on every transport.

Surface contract for unsupported statements:

• CLI and pkg/deltascope: return a partial result with an unsupported array carrying feature and reason fields, plus the ErrUnsupportedStatement sentinel error.
• HTTP and MCP: expose unsupported statements as transport-level errors (HTTP error response, MCP tool error) because the underlying audit function returns an error for unsupported boundaries.

make release-surface-gates VERSION=v0.28.0 and make release-version-surface-gates VERSION=v0.28.0 verify the package/release and versioned docs surfaces.

Previous milestone: v0.26.0 tightened the PostgreSQL CREATE TABLE unsupported boundary contract. See the v0.26.0 release notes for details.

Need PostgreSQL offline audit support?

• Install the normal DeltaScope main archive on supported macOS and Linux platforms; no separate PG-only installer is required.
• deltascope-pg_<version>_linux_amd64.tar.gz remains available only as a legacy compatibility download for older CLI-only workflows during the transition.

The published core archive format is deltascope_<version>_<os>_<arch>.tar.gz. Development-oriented commands are documented under Dev docs.

Release Contract

Every tag produces core archives named deltascope_<version>_<os>_<arch>.tar.gz containing the deltascope, deltascope-server, and deltascope-mcp binaries. The supported darwin/amd64, darwin/arm64, linux/amd64, and linux/arm64 main archives are PG-capable and support PostgreSQL offline across all three binaries. The installer script, Homebrew Cask, and npm MCP launcher all resolve those platform-specific main archives from GitHub Release assets. deltascope-pg_<version>_linux_amd64.tar.gz may still appear as a legacy compatibility download during the transition, but it is no longer part of the primary install story. See the npm package metadata for the current @fanduzi/deltascope-mcp package version.

Quick Start

Audit a risky DML statement:

deltascope audit --sql "delete from users"

Example excerpt:

Verdict: reject
Statements: 1
Blockers: 1
Warnings: 0
Notices: 0

Statement 1: DELETE
- [blocker] dml.where.require: UPDATE and DELETE statements must include a WHERE clause

Audit a CREATE TABLE statement:

deltascope audit --sql "create table tbl_users (id bigint unsigned not null auto_increment comment 'id', created_at datetime not null default current_timestamp comment 'created', updated_at datetime not null default current_timestamp on update current_timestamp comment 'updated', primary key (id)) comment='users' engine=InnoDB default charset=utf8mb4"

Example excerpt:

Verdict: review
Statements: 1
Blockers: 0
Warnings: 1
Notices: 0

Statement 1: CREATE TABLE
- [warning] ddl.column.default.require: column "id" should define a default value

Audit a file:

deltascope audit --file ./migrations/20260328_add_column.sql

Use JSON output for CI or agents:

deltascope audit \
  --sql "create table tbl_users (id bigint unsigned not null auto_increment comment 'id', created_at datetime not null default current_timestamp comment 'created', updated_at datetime not null default current_timestamp on update current_timestamp comment 'updated', primary key (id)) comment='users' engine=InnoDB default charset=utf8mb4" \
  --format json \
  --fail-on warning

Example JSON shape:

{
  "verdict": "review",
  "summary": {
    "statements": 1,
    "blockers": 0,
    "warnings": 1,
    "notices": 0
  },
  "statements": [ ... ],
  "context": {
    "mode": "offline",
    "dialect": "mysql",
    "dialect_source": "default"
  }
}

Audit a PostgreSQL migration with CI-native output:

deltascope audit --dialect postgresql --file ./migrations/20260409_add_index.sql --format github-actions

Audit a PostgreSQL CREATE TABLE statement with named and inline constraints:

deltascope audit \
  --dialect postgresql \
  --sql "create table orders (id bigint primary key, user_id bigint references users(id), amount numeric not null check (amount >= 0), constraint uniq_orders_user unique (user_id), constraint chk_orders_amount check (amount >= 0));"

Audit a PostgreSQL CREATE TABLE with a named foreign key referencing another table:

deltascope audit \
  --dialect postgresql \
  --sql "create table orders (user_id bigint, constraint fk_orders_user foreign key (user_id) references users(id));"

Audit a PostgreSQL phased migration follow-up statement:

deltascope audit \
  --dialect postgresql \
  --sql "alter table users alter column status set default 'active';"

Audit a PostgreSQL constraint lifecycle statement:

deltascope audit \
  --dialect postgresql \
  --sql "alter table users validate constraint chk_amount;"

Generate SARIF output for GitHub Code Scanning:

deltascope audit --file ./migrations.sql --dialect postgresql --format sarif > deltascope.sarif

When SQL looks like PostgreSQL but the dialect is set to MySQL, DeltaScope emits an advisory notice without auto-switching:

deltascope audit --sql "insert into users(id) values (1) returning id;" --format markdown

To audit PostgreSQL SQL explicitly:

deltascope audit --dialect postgresql --sql "insert into users(id) values (1) returning id;"

DML Impact Estimation

For a selective DML such as DELETE FROM users WHERE id = 42, DeltaScope may add an impact object to the statement result. The object is conservative by design and reports estimated_rows, estimated_ratio, risk_level, confidence, source, reason_codes, and optional notes.

{
  "raw_sql": "DELETE FROM users WHERE id = 42",
  "impact": {
    "estimated_rows": 1,
    "estimated_ratio": 0.0001,
    "risk_level": "low",
    "confidence": "high",
    "source": "metadata",
    "reason_codes": ["pk_equality"],
    "notes": ["refined with table statistics"]
  }
}

Offline mode uses SQL shape only. Metadata-aware mode may refine the estimate with read-only table statistics. DeltaScope does not execute the DML and does not run EXPLAIN ANALYZE.

Audit with live metadata (instance-aware rules):

deltascope audit \
  --sql "alter table orders add index idx_status (status)" \
  --host 127.0.0.1 --port 3306 --user root --ask-password --schema app

See all shipped rules:

deltascope rules

Why DeltaScope

SQL mistakes are cheap to catch before they run and expensive after. DeltaScope gives you one consistent engine across local dev, CI, HTTP service, and MCP so the same policy applies everywhere — no per-tool rule duplication, no dialect surprises.

Key Features

• Create-table governance across identifiers, comments, primary keys, audit columns, charset/collation, indexes, and table options.
• Alter-table governance for destructive actions, compatibility checks, existence validation, and merge guidance.
• Object-lifecycle checks for CREATE VIEW, DROP TABLE, and TRUNCATE TABLE.
• DML protections for WHERE, LIMIT, ORDER BY, subqueries, join conditions, bulk insert patterns, denylisted objects, and conservative affected-row impact estimation.
• Stable product surfaces: deltascope CLI, deltascope-server, deltascope-mcp, and pkg/deltascope.
• deltascope-mcp is the official MCP stdio server and exposes audit_sql, describe_rule, list_rules, and get_capabilities.

MCP Quick Start

No install required. The npm launcher fetches and runs the correct deltascope-mcp binary for your platform automatically.

Launcher requirements:

• Node.js 24 or newer
• supported native targets: darwin or linux, amd64 or arm64

Recommended launcher:

claude mcp add --scope user deltascope -- npx -y @fanduzi/deltascope-mcp
codex mcp add deltascope -- npx -y @fanduzi/deltascope-mcp

For raw stdio TOML, native deltascope-mcp, direct connection, connection_ref, proxy setup, and common errors, see Use DeltaScope MCP.

AI Agent Skill

Works in Claude Code, Codex, Cursor, and 40+ AI coding agents. Install once, get inline SQL review in every session.

DeltaScope ships a universal AI agent skill for inline SQL review during AI coding sessions. The skill detects whether DeltaScope is installed locally, calls it to audit your SQL, and surfaces findings with fix suggestions — without leaving your AI coding session.

# Install via npx skills (Claude Code, Codex, Cursor and 40+ AI agents)
npx skills add Fanduzi/DeltaScope --skill deltascope-review -a claude-code

Install globally (available across all projects):

npx skills add Fanduzi/DeltaScope --skill deltascope-review -a claude-code -g

Keep the skill up to date:

npx skills update

Then invoke in any supported AI session:

/deltascope-review

Paste a SQL snippet or point to a file — the agent audits it with DeltaScope and suggests fixes. See skills/README.md for full setup and usage.

More Docs

• Recipes
• Dev docs
• Reference docs
• Audit SQL with metadata
• Review DDL before migration
• Guard DML in CI
• Use with AI agents
• Inspect rules and config
• Troubleshoot metadata-aware audit

Documentation

• Admin docs
• Concept docs
• Dev docs
• Reference docs
• Audit capability matrix

Developer Workflows

• make test runs go test ./...
• make build produces all local binaries under bin/
• make build-linux produces Linux amd64 binaries under bin/
• make test-e2e-cli runs the Docker-backed metadata CLI smoke suite
• make pg-unit-test-gates runs the PostgreSQL-tagged unit gate set
• make pg-e2e-gates runs the Docker-backed PostgreSQL CLI, HTTP, and MCP suites
• make pg-confidence-gates runs the canonical PostgreSQL confidence closure
• docs/dev/testing.md covers the full target set

HTTP Service

Run the HTTP adapter over the same audit engine:

deltascope-server -listen 127.0.0.1:8083

Endpoints:

• GET /healthz
• GET /version
• POST /v1/audit

POST /v1/audit supports both offline JSON audit requests and metadata-aware requests with an optional connection block. The HTTP response keeps the public audit result body and adds a context block. See the full contract in HTTP API reference.

Library Usage

result, err := deltascope.Audit(ctx, deltascope.Request{
    SQL:     "delete from users",
    Dialect: deltascope.DialectMySQL,
})

The stable public API lives in pkg/deltascope.

Architecture

DeltaScope keeps one audit path and exposes it through multiple entrypoints. Product-level and implementation-level diagrams live in docs/concept/architecture.md and docs/dev/architecture.md.

Modules

Module	Description	Doc
cmd/deltascope	CLI process entrypoint	README
cmd/deltascope-server	HTTP service entrypoint	README
cmd/deltascope-mcp	MCP service entrypoint	README
internal/interfaces	Transport adapter namespace	README
internal/interfaces/cli	CLI adapter layer	README
internal/interfaces/http	HTTP adapter layer	README
internal/interfaces/mcp	MCP adapter layer	README
internal/application	Use-case orchestration layer	README
internal/application/audit	Application parse/audit orchestration	README
internal/application/auditmeta	Shared metadata-aware audit preparation	README
internal/application/policy	Application policy loader	README
internal/domain	Core domain types and rules	README
internal/domain/spec	Normalized statement specifications	README
internal/domain/rule	Rule findings and severity model	README
internal/domain/rule/catalog	Explanation-oriented shipped rule catalog	README
internal/domain/rule/ddl	DDL rule catalog	README
internal/domain/rule/dml	DML rule catalog	README
internal/domain/policy	Policy configuration model	README
internal/domain/report	Audit result aggregation and verdicts	README
internal/infrastructure	Infrastructure adapter layer	README
internal/infrastructure/parser	Parser adapter namespace	README
internal/infrastructure/parser/tidb	TiDB parser adapter	README
internal/infrastructure/config/viper	YAML config adapter	README
internal/infrastructure/metadata/mysql	Metadata provider for MySQL/TiDB-compatible engines	README
internal/infrastructure/output	Output renderer namespace	README
internal/infrastructure/output/markdown	Markdown renderer	README
internal/infrastructure/output/json	JSON renderer	README
configs	Example configuration files	README
pkg/deltascope	Stable public package surface	README