4d9a26498e
Paste a secret, get a link, send it. The first person to open it and press
Reveal sees the secret; the link dies at that moment. The recipient needs a
browser and nothing else — no account, no client, no installed tooling.
The server cannot read what it stores. AES-256-GCM happens in the browser and
the key lives in the URL fragment, which browsers never transmit, so hushd
holds ciphertext and no key material. That is a property of where the key sits
rather than a promise about our conduct, which is why there is deliberately no
endpoint accepting a plaintext secret and no server-side-encryption fallback:
two guarantees behind one URL would be worse than one honest guarantee.
Three decisions carry the design:
* GET /s/{id} touches NO storage, not even to check existence. Slack, Teams,
WhatsApp, iMessage and Outlook Safe Links all fetch a URL before a human
sees it, so destroying on GET would destroy most secrets in transit and the
recipient's "already used" would be indistinguishable from interception.
Only POST /reveal consumes. Bot user-agent detection is an arms race;
removing the side effect from GET is not. Pinned by
TestGettingTheRevealPageNeverConsumesTheSecret.
* Destruction is one Redis GETDEL, which is atomic. GET-then-DEL has a window
where two simultaneous readers both win, and for a one-time secret that
window is the product. The store contract demands atomicity and the same
concurrency test runs against both implementations.
* Missing, already-revealed, expired and evicted are ONE indistinguishable
410. Separating them would confirm to a prober that a given link was real.
The secret id IS the capability, so secret.ID is a struct whose every
accidental path — %v, %s, String(), slog, json.Marshal — emits a redacted
handle or refuses, and the raw value needs an explicit Value(). The first
version tried to prevent leaks by implementing no String() at all; its own test
caught that Go's fmt prints unexported fields anyway, so forbidding the method
had removed the control rather than the leak.
Operationally: structured JSON on stdout in the fleet's wire format, which
Vector already collects with no annotation; six hush_* metrics on the chassis
registry with no id, IP or path in any label; five alert rules wired into
vmalert. The public Ingress enumerates /, /s/ and /api/ so /metrics, /healthz
and /readyz share the port but are unreachable from the internet — no
basic-auth middleware to maintain and get wrong.
Dependencies are vendored because go-chassis is private: the Woodpecker test
step and the in-cluster Kaniko build both run -mod=vendor with GOPROXY=off and
hold no git credential.
cmd/hush-mcp is a stdio MCP server doing the same client-side crypto locally,
so using hush from an agent preserves the same guarantee as using it from a
browser.
81 lines
3.3 KiB
Go
81 lines
3.3 KiB
Go
package secret
|
|
|
|
import (
|
|
"encoding/base64"
|
|
"errors"
|
|
"fmt"
|
|
"time"
|
|
)
|
|
|
|
// Size and lifetime policy. Every bound here is a product decision with a
|
|
// reason, not a tuning knob:
|
|
//
|
|
// - MaxCiphertextBytes keeps hush a courier for credentials rather than a file
|
|
// host. 64 KiB of AES-GCM holds roughly 48 KiB of plaintext, which is a very
|
|
// large credential and a very small file.
|
|
// - MinTTL exists because a link that expires before the recipient reads their
|
|
// messages is a support ticket, not a security win.
|
|
// - MaxTTL bounds exposure. The residual risk in this design is the key sitting
|
|
// in a browser history entry, and a week is as long as that is defensible.
|
|
const (
|
|
MaxCiphertextBytes = 64 * 1024
|
|
MinTTL = 5 * time.Minute
|
|
MaxTTL = 7 * 24 * time.Hour
|
|
DefaultTTL = 24 * time.Hour
|
|
)
|
|
|
|
// Validation failures. Each maps to one API error code, so a caller can branch
|
|
// on the cause without parsing prose.
|
|
var (
|
|
ErrCiphertextEmpty = errors.New("ciphertext is empty")
|
|
ErrCiphertextTooLarge = errors.New("ciphertext exceeds the size limit")
|
|
ErrCiphertextInvalid = errors.New("ciphertext is not valid base64url")
|
|
ErrTTLOutOfRange = errors.New("ttl is outside the permitted range")
|
|
)
|
|
|
|
// ciphertextEncoding matches what the browser produces: base64url, unpadded.
|
|
var ciphertextEncoding = base64.RawURLEncoding
|
|
|
|
// ValidateCiphertext checks what the server is ABLE to check. hush cannot
|
|
// verify that the bytes decrypt, because it has no key — by design. So it
|
|
// verifies the two things it can: that the encoding is what this service's
|
|
// clients produce, and that the size is inside the cap.
|
|
//
|
|
// The encoding check is not cosmetic. Without it, hush becomes a store for
|
|
// arbitrary bytes addressable by URL, which is a different and much less
|
|
// defensible service than the one described in the README.
|
|
func ValidateCiphertext(ciphertext string) error {
|
|
switch {
|
|
case ciphertext == "":
|
|
return ErrCiphertextEmpty
|
|
case len(ciphertext) > MaxCiphertextBytes:
|
|
// Measured on the encoded form, which is what is stored and what
|
|
// bounds memory. Checked BEFORE decoding so an oversized body is
|
|
// rejected without allocating its decoded copy.
|
|
return fmt.Errorf("%w: %d > %d bytes", ErrCiphertextTooLarge, len(ciphertext), MaxCiphertextBytes)
|
|
}
|
|
if _, err := ciphertextEncoding.DecodeString(ciphertext); err != nil {
|
|
return ErrCiphertextInvalid
|
|
}
|
|
return nil
|
|
}
|
|
|
|
// ResolveTTL turns a caller's requested lifetime into the one that will be
|
|
// used. Zero means "unspecified" and gets the default.
|
|
//
|
|
// An out-of-range value is an ERROR, never a silent clamp. A caller who asked
|
|
// for 30 days and got 7 without being told would believe their link outlives
|
|
// its actual expiry, and would find out when the recipient could not open it.
|
|
func ResolveTTL(requested time.Duration) (time.Duration, error) {
|
|
if requested == 0 {
|
|
return DefaultTTL, nil
|
|
}
|
|
if requested < MinTTL || requested > MaxTTL {
|
|
return 0, fmt.Errorf("%w: %s not in [%s, %s]", ErrTTLOutOfRange, requested, MinTTL, MaxTTL)
|
|
}
|
|
// Truncate to whole seconds: the wire format is seconds and Redis EX takes
|
|
// seconds, so keeping sub-second precision would make the expires_at we
|
|
// report disagree with the expiry Redis enforces.
|
|
return requested.Truncate(time.Second), nil
|
|
}
|