The Silo Manifesto
Silo exists for one purpose: to keep a trustworthy, maintained, feature-complete community edition of S3-compatible object storage in existence, and to keep its software supply chain unbroken.
This page is the project’s public commitment. It follows one rule throughout: every article is either something we already practice with public evidence, or something we deliberately refuse to promise. A commitment we could not keep would be worse than none.
Article 1 · Reason to exist, and an exit clause
This project was born when upstream wound down its community edition: the web console was stripped, prebuilt community binaries stopped, and the community repository was archived. Silo provides continuity for existing MinIO-compatible deployments — Pigsty runs it in production as its PostgreSQL backup storage, so our own operations depend on the same artifacts we publish.
The fork is a means, not an identity. If upstream restores its commitment to a community edition, we will welcome it, gladly narrow our scope, and offer our fixes back.
Background:
- MinIO is Dead (2025-12) — what upstream removed, and when.
- MinIO Is Dead. Which Next? (2025-12) — the alternatives, evaluated.
- MinIO Is Dead, Long Live MinIO (2026-02) — the fork, announced.
- MinIO Fork, Promise Kept (2026-04) — the first months of receipts.
Article 2 · The compatibility contract
The product and its trademark are renamed; the protocol and your data are not.
- The S3 API,
MINIO_*environment variables,minio_*metrics,x-minio-*headers,/minio/*routes, and the on-disk format (including.minio.sys) are preserved, and frozen by a CI compatibility guard. - Every release documents its tested rollback target and path in the release notes.
- New capabilities do not touch the on-disk format. Any exception must be explicitly marked as non-reversible before you can enable it.
- The migration guide — including how to leave Silo — and the per-component compatibility audits are public and maintained.
Article 3 · The license is constant, structurally
Silo is AGPLv3, forever. There is no CLA and no copyright aggregation: contributions are accepted only with DCO sign-off, so copyright stays with each contributor. Not relicensing is not a verbal promise — we have made it structurally impossible.
Our reading of the AGPL boundary is itself a public commitment, stated as a position rather than legal advice: using Silo through its S3 API does not make your application a derivative work. The license will never be used as a threat or a sales instrument.
Article 4 · Change discipline
Changes relative to the upstream baseline fall into four classes only: security fixes, defect fixes, restored community features, and optional additions. Existing API semantics change only when a security fix demands it, and every such compatibility cost is recorded in the advisory that caused it. The full divergence from upstream is maintained as a code-verified compatibility audit.
Article 5 · The never list
Silo will never:
- move an existing feature behind a paywall;
- put a registration or login wall in front of downloads;
- ship telemetry — the upstream phone-home paths (update checks, SUBNET, call-home) are removed outright, not merely disabled;
- require a CLA;
- change the license;
- use trademarks against normal use or descriptive mention.
This list is append-only: entries may be added, never removed.
Article 6 · Security discipline
- Vulnerabilities are reported through a private channel, triaged reproducibly, and disclosed in coordination — writeups are held until the fixed release ships.
- Every security fix is published with a public advisory in the security chronicle, including its evidence and its compatibility cost.
- Dependencies are tracked for CVEs, with reachability analysis before any change.
- Every release ships SHA-256 checksums, SPDX SBOMs, Sigstore-signed manifests, and GitHub build provenance.
Severe, remotely exploitable issues are expedited on a best-effort basis, but we do not commit to a specific fix-time SLA.
Article 7 · Release rhythm
Releases ship typically every one to two months, and at most a quarter apart. Security and defect fixes are batched into these releases. The history is public; judge the promise against it.
Version tags keep the RELEASE.YYYY-MM-DDTHH-MM-SSZ format, and each release documents its upstream baseline.
Deprecations get at least six months’ notice with a migration path — except where security requires immediate removal, in which case the removal and its justification are published in the security chronicle.
Article 8 · Upstream relations
We renamed the project out of respect for upstream’s trademarks. Upstream copyright, license, and third-party notices are preserved in full: attribution and trademark. If upstream resumes accepting contributions, applicable fixes will be offered back as appropriate.
Article 9 · Continuity
- The repositories live under the pgsty organization, not a personal account.
- The build is documented and provenance-attested: anyone can rebuild equivalent artifacts from source without us.
- If active maintenance stops for six months, we will say so publicly and archive the project tidily — and make our best effort to keep released artifacts and documentation available.
- Should an established open-source organization — a CNCF-style foundation, for example — want to bring Silo under more formal governance, we would be glad to cooperate.
Article 10 · The commercial boundary
Everything in the Silo repositories — the server, the client, the console, and the released artifacts — is complete and free of charge, now and in the future.
Article 11 · Amendment discipline
Additions and strengthenings of this manifesto take effect immediately. Weakening or removing any article requires ninety days’ public notice. Article 5 is append-only, always.
The ongoing evidence:
- Security chronicle: every investigated CVE and fix, one article per incident.
- Release notes: every release, with its baseline, rollback target, and acceptance record.
- Compatibility audits: where Silo matches MinIO, and where it deliberately differs.