Files
beardthelion 6edf9a8b78 fix(sandbox): normalize launcher names before the command-prefix denylist (#934)
* fix(sandbox): retire recorded approvals through a policy bump

Ahead of tightening which command prefixes may be approved, make the
upgrade path safe. A grant already on disk that the narrower rules
reject fails validation on read, and that path rejects the entire grants
file rather than the one entry. Every caller treats a read error as "no
grant" (engine.go LookupCommandPrefix, ApprovedCommandPrefixes, and the
Lookup in Evaluate), so one stale prefix would silently disable the
user's persisted deny grants too, with the error text never reaching
them.

Bump the approval policy version instead and let the store's existing
migration handle it: it backs the file up, keeps deny grants,
invalidates prior approvals, and reports the counts once through the
startup notice that both frontends already print. That is what the
mechanism was built for.

The v1-migration test pinned the policy version as a literal, so it now
derives the expected value from the constant and will not need editing
on the next bump.

* fix(sandbox): normalize launcher names before the command-prefix denylist

unsafeCommandPrefixLauncher matched the program name as written, so
python3.11, python2.7, python.exe, node.exe and git.exe all validated as
ordinary commands and could be persisted as command-prefix grants.
Lookup is a string-prefix match, so a single approval then auto-allowed
every later invocation: granting [python3.11 -c] covers
`python3.11 -c "import os; os.system(...)"` with no further prompt.

internal/agent's commandName already normalizes case and Windows
executable extensions for the allow side of the same decision; the deny
side did not, and the two disagreeing is the bug. Reduce the program to
the launcher it actually runs and match on that: executable extension,
Windows' trailing dots and spaces (python. starts python.exe), CPython
ABI flags, then a trailing version, testing the digit-stripped form as
well so python3.11 and python3 both arrive as python. A name is only
ever narrowed toward an existing entry, so the failure direction is an
extra permission prompt, never a silent grant. Digits without a
separator stay part of the name, which keeps base64, 7z and sha256sum
grantable.

Two Windows spellings cannot be resolved by name at all, so the shape is
refused alongside a path: 8.3 short names truncate the stem, which makes
POWERS~1.EXE unrecognizable, and python.exe::$DATA reaches the same
executable through its default stream. Normalizing the first token
before the banned-suggestion comparison keeps git.exe at parity with
git.

The list also gains launchers of the same class it already covers and
that normalization cannot reach: cmd, wsl, the remaining POSIX shells,
busybox, uv and uvx. Entries normalization now covers (/bin/bash,
/bin/zsh, powershell.exe, python3, pypy3) are gone; the first two were
already unreachable behind the path check above them.

Covers the persisted store, the session-scoped grants the agent takes
during a turn, the legacy-schema migration that prunes such a grant
per entry, and the Windows spellings hermetically.

* fix(sandbox): treat free-threaded CPython as a launcher too

The ABI flags came from the pre-3.13 set, so "python3.13t" and
"python3.13t.exe" normalized to themselves and stayed grantable as
command prefixes while the GIL build "python3.13" was refused. A
free-threaded interpreter runs arbitrary code just the same.

Add "t" to the flags trimmed under the existing digit guard, which is
what keeps ordinary names intact: "zstd" and "cat" have no digit beneath
the trimmed letters, so they keep their own names and stay grantable.

Reported by CodeRabbit on #934.

* fix(sandbox): refuse the remaining spellings of listed launchers

Same defect as the free-threaded build, found by sweeping the families
already on the list for names it cannot reach:

- nodejs is node on Debian and Ubuntu, and was grantable.
- python3-dbg, python3.11-dbg and python3.13t-dbg are interpreters that
  run arbitrary code; so are bash-static, pwsh-preview and
  node-nightly. A closed set of build-channel suffixes now normalizes
  them. The set stays closed deliberately: a general "-word" strip would
  also swallow node-gyp, python3-config and ruby-lsp, which are ordinary
  tools a user should still be able to approve, and those are pinned as
  must-stay-grantable.
- pnpm, yarn and bunx run package scripts exactly as npm and npx do, and
  npm and npx are already refused.
- sudoedit escalates like sudo.

The launcher-wrapper family (winpty, chroot, unshare, nsenter, parallel,
script, unbuffer, taskset, flock) has the same inconsistency against the
env/nohup/timeout/setsid entries already on the list, and is left for a
separate decision rather than widened here.

* test(sandbox): cover the -debug and -beta launcher suffixes

Four of the six build-channel suffixes had regression cases; -debug and
-beta did not, so removing either from the set left every test green.
Both now fail RED when their entry is dropped.

Reported by CodeRabbit on #934.
2026-08-21 12:50:11 +08:00

258 lines
8.7 KiB
Go

// Package execution defines the platform-neutral command execution contract.
// Frontends and tools use these types without depending on native sandbox or
// process-management details.
package execution
import (
"errors"
"fmt"
"strings"
)
// PolicyVersion is the version of the approval policy that recorded grants were
// evaluated under. Bump it whenever a change narrows what may be approved: the
// grant store then backs the file up, keeps deny grants, invalidates prior
// approvals, and tells the user once, instead of silently honoring approvals
// that the current rules would have refused.
//
// v2 tightened command-prefix validation so a launcher spelled with a version
// or an executable extension ("python3.11", "python.exe") can no longer be
// approved as an ordinary command.
const PolicyVersion = 2
type Origin string
const (
OriginInteractiveCommand Origin = "interactive_command"
OriginHook Origin = "hook"
OriginPlugin Origin = "plugin"
OriginMCPServer Origin = "mcp_server"
OriginBackgroundTask Origin = "background_task"
)
type Mode string
const (
ModeCaptured Mode = "captured"
ModeInteractive Mode = "interactive"
ModeDurable Mode = "durable"
)
type Command struct {
Name string `json:"name"`
Args []string `json:"args,omitempty"`
Env []string `json:"env,omitempty"`
}
type CapabilityKind string
const (
CapabilityRead CapabilityKind = "filesystem_read"
CapabilityWorkspaceWrite CapabilityKind = "workspace_write"
CapabilityProtectedMetadata CapabilityKind = "protected_metadata"
CapabilityExternalNetwork CapabilityKind = "external_network"
CapabilityIsolatedLoopback CapabilityKind = "isolated_loopback"
CapabilityHostVisibleBinding CapabilityKind = "host_visible_binding"
CapabilityProcessSpawn CapabilityKind = "process_spawn"
CapabilityProcessControl CapabilityKind = "process_control"
CapabilityEnvironment CapabilityKind = "environment"
CapabilityUnrestricted CapabilityKind = "unrestricted"
)
type Capability struct {
Kind CapabilityKind `json:"kind"`
Scope string `json:"scope,omitempty"`
}
type ApprovalContext struct {
SessionID string `json:"sessionId,omitempty"`
PolicyVersion int `json:"policyVersion"`
}
type Request struct {
Origin Origin `json:"origin"`
Mode Mode `json:"mode"`
Command Command `json:"command"`
WorkingDirectory string `json:"workingDirectory"`
WorkspaceRoots []string `json:"workspaceRoots"`
Capabilities []Capability `json:"capabilities,omitempty"`
Approval ApprovalContext `json:"approval"`
}
func (request Request) Validate() error {
if !validOrigin(request.Origin) {
return errors.New("execution request requires a valid origin")
}
if request.Mode != ModeCaptured && request.Mode != ModeInteractive && request.Mode != ModeDurable {
return errors.New("execution request requires a valid mode")
}
if strings.TrimSpace(request.Command.Name) == "" {
return errors.New("execution request requires a command")
}
if strings.TrimSpace(request.WorkingDirectory) == "" {
return errors.New("execution request requires a working directory")
}
if len(request.WorkspaceRoots) == 0 {
return errors.New("execution request requires at least one workspace root")
}
for _, root := range request.WorkspaceRoots {
if strings.TrimSpace(root) == "" {
return errors.New("execution request contains an empty workspace root")
}
}
for _, capability := range request.Capabilities {
if !validCapabilityKind(capability.Kind) {
return fmt.Errorf("execution request contains invalid capability %q", capability.Kind)
}
}
return nil
}
type State string
const (
StateRequested State = "requested"
StateEvaluated State = "evaluated"
StateAwaitingApproval State = "awaiting_approval"
StatePrepared State = "prepared"
StateRunning State = "running"
StateRetained State = "retained"
StateCompleted State = "completed"
StateDenied State = "denied"
StateFailed State = "failed"
StateCancelled State = "cancelled"
)
type OutcomeKind string
const (
OutcomeRunning OutcomeKind = "running"
OutcomeSuccess OutcomeKind = "success"
OutcomePolicyDenied OutcomeKind = "policy_denied"
OutcomeEnforcementDenied OutcomeKind = "enforcement_denied"
OutcomeApplicationFailure OutcomeKind = "application_failure"
OutcomeSandboxSetupFailure OutcomeKind = "sandbox_setup_failure"
OutcomeExecutableNotFound OutcomeKind = "executable_not_found"
OutcomeCancelled OutcomeKind = "cancelled"
OutcomeTimedOut OutcomeKind = "timed_out"
OutcomeEnforcementDegraded OutcomeKind = "enforcement_degraded"
)
type Exit struct {
Code int `json:"code"`
Signal string `json:"signal,omitempty"`
}
type DenialSource string
const (
DenialSourceConfiguredPolicy DenialSource = "configured_policy"
DenialSourcePlatformSandbox DenialSource = "platform_sandbox"
DenialSourceApproval DenialSource = "approval"
)
type DenialNextAction string
const (
DenialNextActionNone DenialNextAction = "none"
DenialNextActionRequestApproval DenialNextAction = "request_approval"
DenialNextActionChangeCommand DenialNextAction = "change_command"
DenialNextActionEnableSandbox DenialNextAction = "enable_sandbox"
)
type Denial struct {
Capability Capability `json:"capability"`
Source DenialSource `json:"source"`
Reason string `json:"reason"`
Recoverable bool `json:"recoverable"`
NextAction DenialNextAction `json:"nextAction"`
}
type Enforcement struct {
Backend string `json:"backend,omitempty"`
Level string `json:"level,omitempty"`
Degraded bool `json:"degraded,omitempty"`
DowngradeReason string `json:"downgradeReason,omitempty"`
}
type Outcome struct {
State State `json:"state"`
Kind OutcomeKind `json:"kind"`
ProcessID string `json:"processId,omitempty"`
Exit *Exit `json:"exit,omitempty"`
Denial *Denial `json:"denial,omitempty"`
Enforcement Enforcement `json:"enforcement"`
Changes []Change `json:"changes,omitempty"`
}
// AdapterReport is the structured, machine-readable result emitted by a
// platform enforcement adapter. It is separate from stdout and stderr so
// command text cannot impersonate a policy decision.
type AdapterReport struct {
Denial *Denial `json:"denial,omitempty"`
}
func (outcome Outcome) Validate() error {
if outcome.State == "" || outcome.Kind == "" {
return errors.New("execution outcome requires state and kind")
}
switch outcome.State {
case StateRetained, StateRunning:
if outcome.Kind != OutcomeRunning || strings.TrimSpace(outcome.ProcessID) == "" {
return errors.New("running execution outcome requires a process id")
}
case StateCompleted:
if outcome.Kind != OutcomeSuccess || outcome.Exit == nil || outcome.Exit.Code != 0 {
return errors.New("completed execution outcome requires a successful exit")
}
case StateDenied:
if (outcome.Kind != OutcomePolicyDenied && outcome.Kind != OutcomeEnforcementDenied) || outcome.Denial == nil {
return errors.New("denied execution outcome requires structured denial details")
}
if err := outcome.Denial.Validate(); err != nil {
return err
}
case StateFailed:
if outcome.Kind == OutcomeRunning || outcome.Kind == OutcomeSuccess || outcome.Exit == nil {
return errors.New("failed execution outcome requires a failure kind and exit")
}
case StateCancelled:
if outcome.Kind != OutcomeCancelled {
return errors.New("cancelled execution outcome requires cancelled kind")
}
default:
return fmt.Errorf("execution outcome state %q is not terminal or externally observable", outcome.State)
}
return nil
}
func (denial Denial) Validate() error {
if !validCapabilityKind(denial.Capability.Kind) {
return errors.New("execution denial requires a valid capability")
}
if denial.Source == "" || strings.TrimSpace(denial.Reason) == "" || denial.NextAction == "" {
return errors.New("execution denial requires source, reason, and next action")
}
return nil
}
func validOrigin(origin Origin) bool {
switch origin {
case OriginInteractiveCommand, OriginHook, OriginPlugin, OriginMCPServer, OriginBackgroundTask:
return true
default:
return false
}
}
func validCapabilityKind(kind CapabilityKind) bool {
switch kind {
case CapabilityRead, CapabilityWorkspaceWrite, CapabilityProtectedMetadata,
CapabilityExternalNetwork, CapabilityIsolatedLoopback, CapabilityHostVisibleBinding,
CapabilityProcessSpawn, CapabilityProcessControl, CapabilityEnvironment, CapabilityUnrestricted:
return true
default:
return false
}
}