# Forge Tenancy

> Verify tenant context propagation and isolation across data, cache, files, jobs, search, analytics, and administration.

- Skill: `is-bo/forge-tenancy-3` (Agent Skill)
- Install (CLI): `npx skillmds@latest add is-bo/forge-tenancy-3`
- Raw SKILL.md: https://api.skillmd.com/api/skills/is-bo/forge-tenancy-3/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Data & Analytics
- Author: is-bo (https://skillmd.com/u/is-bo)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/is-bo/forge-tenancy-3

---


# forge-tenancy: Multi-tenancy

Engine: Forge native

## Purpose

Verify tenant context propagation and isolation across data, cache, files, jobs, search, analytics, and administration.


## Deterministic runtime composition

Before loading any provider procedure, run:

Resolve `../../runtime/cli/src/composition-entry.js` relative to this `SKILL.md`, then run:

`node "<resolved-absolute-runner-path>" tenancy compose --workflow audit --root "<repository-root>" --dry-run --json`

Add one repeatable `--request <provider-or-source>` flag for each explicit user request. Add
`--condition <task-condition>` or `--risk-surface <surface>` only for a task fact you directly
proved; never infer one from generic wording. The command above is the default for this
audit-oriented module; for implementation use `--workflow build`, and for a fix, retest, or
release gate use `--workflow fix`, `verify`, or `ship` respectively. Read the JSON response,
keep the Forge contract at index zero, and resolve paths against the absolute `runtime_root`
reported in that response. Read `eager[].runtimePath` when entering the module. The full
`selected[]` list is availability/provenance; load only `deferred[].runtimePath` when the task
reaches that concern, in tier order. Refuse any path that escapes the root. Respect every reported
suppression and context budget. If `missing` is non-empty, stop and report the installation as
damaged; do not improvise a prose fallback. The runner and specialist content may live in a plugin
cache or global installation; never assume they are inside the audited repository.


Resolve and read `../fullstack-forge/references/shared/module-contract.md` (applicability,
execution, mutation, verification, completion) and
`../fullstack-forge/references/shared/evidence-rules.md` (statuses, standards, tools, findings via
`../fullstack-forge/references/PROTOCOL.md`) relative to this module `SKILL.md` before reporting.

Never hide failed checks or claim that an operation ran when it did not.

## Automatic activation signals

Activate when a request or direct repository evidence involves multi-tenancy, when
the user explicitly names `forge-tenancy`, or when discovery proves an applicable boundary.

- Shared applications serving distinct organizations or customer partitions

## When not to activate

- Single-tenant deployments with no shared data plane

## Automated support

Relevant discovery inputs are:

- tenant model
- authorization policies
- queries, cache keys, storage paths, and jobs

Deterministic support, bounded evidence only:

- `inspect-authorization`
- `inspect-database-schema`
- `inspect-cache-usage`

## Agent inspection procedure

1. Locate the tenant identifier's origin and verify it derives from authenticated identity, never from request input.
2. Trace tenant-context propagation through every data query, verifying the tenant predicate at the final data-access boundary.
3. Check isolation beyond the database: cache keys, file paths, background jobs, logs, metrics, and search indexes.
4. Inspect unique constraints and identifiers for tenant scoping, and admin or support access for explicit cross-tenant audit trails.
5. Run or demand negative tests: the same object identifier across two tenants on every access path, including exports and downloads.

Manual inspection requirements:

- Review provisioning, suspension, migration, deletion, and break-glass flows
- Validate database and provider isolation settings

Stack-specific guidance:

- Ensure ORM helpers cannot be bypassed by raw queries or background workers

## Evidence to collect

Standards used as criteria:

- OWASP Multi Tenant Security Cheat Sheet
- OWASP Authorization Cheat Sheet

## Common production failures

- Identify the trusted tenant-context source and follow it through every boundary
- Test missing, forged, stale, and cross-tenant identifiers in reads, writes, bulk actions, exports, jobs, and websockets
- Inspect composite uniqueness, row policies, cache namespaces, object prefixes, logs, and support tooling

## Missing-control checks

Each item needs direct evidence or one reasoned status.

- Tenant identifiers
- Tenant propagation
- Query scoping
- Cache-key isolation
- File-path isolation
- Object-storage isolation
- Background jobs
- Scheduled jobs
- Logs
- Metrics
- Unique constraints
- Exports
- Admin access
- Billing ownership
- Tenant deletion
- Shared resources
- Tenant-specific secrets
- Tenant-specific rate limits
- Cross-tenant testing
- Active negative tests for cross-tenant access

## Commands and tools

- Run `forge tenancy audit --json` or `fullstack-forge tenancy audit --json` when
  an explicit audit is requested and the CLI is installed. Normal feature work does not require it.

## Safe fixes

- Add tenant-scoped tests and explicit cache-key namespaces
- Reject missing tenant context at an existing boundary

## Approval-required changes

- Changing tenant identifiers, isolation strategy, or support access

## Verification

- Run cross-tenant negative tests for every critical resource
- Confirm asynchronous work restores trusted tenant context

## Completion contract

Follow `fullstack-forge/references/shared/completion.md` and the limitations below.

## Known limitations

- Infrastructure-level isolation requires live configuration evidence