Files
marqdo-cli/spec/security.md
T
cflmy 2c7c4f31e7
Linux strict security lane (G-C) / strict isolation escape matrix (push) Waiting to run
ci/woodpecker/push/main Pipeline was successful
ci/woodpecker/manual/performance Pipeline was successful
CI / fmt + clippy + test (Linux x86_64) (push) Waiting to run
CI / schema instance validation + consumers (push) Waiting to run
CI / runtime contract (macos-latest) (push) Waiting to run
CI / runtime contract (windows-latest) (push) Waiting to run
release(3.0.0-beta.1): product version migration, beta gates, archive tooling and developer docs
REL-01: workspace/CLI/Analysis product version 3.0.0-beta.1 with updated
Cargo.lock; project.marqdo is enforced against the running core with
package.marqdo_range_invalid/package.marqdo_incompatible fail-closed
diagnostics; Tool SemVer prerelease negotiation tests; init template and
examples use the beta range. Verify init derives the project name from the
target directory instead of falling back to 'example'.

REL-02: beta support matrix is Linux x86_64 only; macOS/Windows are explicitly
experimental/unclaimed; strict policy and G4 Beta-only WARN are documented for
the release evidence.

REL-03: deterministic release packaging (tools/release/package.py) ships the
CLI and provider worker together with SOURCE_SHA, RELEASE_INFO.json,
known-limitations, specs, developer docs, examples incl. built plugin/provider
artifacts and consumers; tools/release/smoke.py cold-install tests the archive.

REL-04: docs/developers entry pages; corrected and self-contained third-party
integration handbook; stale Proposed/native-only/informational spec claims
fixed; release/status/compatibility/changelog/releasing docs updated.

REL-05: release_evidence.py gains an explicit beta profile
(BETA_RELEASE_CANDIDATE) distinct from the stable profile; Woodpecker build
step now packages and smoke-tests the archive on every pipeline.

Adds a runnable Resource Provider v2 example and corrects check --emit to
Analysis v2.
2026-11-22 02:31:32 +08:00

26 KiB

Marqdo Security Model (implemented status)

Status: describes the 3.0.0-beta.1 implementation, including honest gaps. This is not a security audit; it documents what the code actually enforces.

Normative implementation:

Concern Code
Sandbox modes, strict refusal crates/marqdo-sandbox/src/lib.rs
Capability sets and checks crates/marqdo-host/src/capability.rs
Trusted capability handlers crates/marqdo-host/src/capability.rs (CapabilityBroker, CapabilityHandler, FsReadHandler; ADR-0057)
Host dispatch and per-op checks crates/marqdo-host/src/host.rs
Project-rooted filesystem crates/marqdo-host/src/fs.rs
Budgets crates/marqdo-host/src/budget.rs
Effect tracing crates/marqdo-host/src/trace.rs
Math/foreign adapters crates/marqdo-host/src/kernels.rs
Foreign child processes crates/marqdo-foreign/src/lib.rs
Plugin loading / in-process plugin calls crates/marqdo-host/src/host.rs (load_plugin, op_plugin_invoke), crates/marqdo-plugin-abi/src/loader.rs
CLI flags and capability construction crates/cli/src/main.rs, crates/cli/src/commands.rs

1. Modes

SandboxMode has three values (pure, project, strict), parsed by SandboxMode::parse and selectable on run/call with --mode <pure|project|strict> (default: project).

Mode Intended guarantee Actual status in this build
pure No external I/O; memory computation and budgets only. Enforced. Capabilities::pure() = console, math only; every FS/foreign/plugin grant flag makes the CLI exit 2 with "pure mode prohibits external I/O grants".
project Operator-approved project resources under CapabilityHost + bounded resources. Enforced for the implemented operations. Defaults equal pure; --allow-read, --allow-write, --allow-net, --allow-env, --allow-process, --allow-random, --allow-foreign, --allow-plugin add grants. resource.open supports stdin read, separate stdout/stderr write, granted project-file read/write (ADR-0025/0026/0040), granted in-process provider kinds (ADR-0033) and granted TCP/UDP/listener/TLS streams, including DNS names after an exact host grant (ADR-0034/0047/0048); process/device remain unsupported. Sandboxing is cooperative (in-process checks), not OS isolation; in-process providers are trusted, not sandboxed.
strict Untrusted documents under real OS isolation. Linux: enforced when the runtime probe passes (ADR-0021). resource.open is refused in strict except for the existing stdio/stdout console channel (console grant); stdin/stderr, file/TCP/provider/process resources remain refused until a provider passes the isolation lane (CM-A11, ADR-0030/0034/0040). PlatformCapabilities::detect() runs a real Bubblewrap smoke test; success reports Available { mechanism }, failure reports Unavailable { reason }. Strict L1 supports the interpreter, console, native Math and the read-only bundled stdlib (published at /run/marqdo-stdlib for the Compat-C print chain); --allow-read/--allow-write/--allow-net/--allow-env/--allow-process/--allow-random/--allow-foreign/--allow-plugin/--plugin/--trust-native-plugin/--kernel/--trace-file are rejected with exit 2. macOS/Windows and Linux hosts without the mechanisms remain Unavailable; there is no fallback to an unrestricted process.

marqdo inspect <file> --platform prints the detection report, for example:

{
  "os": "linux",
  "arch": "x86_64",
  "strict": {"state": "available",
             "mechanism": "bubblewrap 0.11.1 (user/mount/PID/net/UTS/IPC namespaces + no-new-privs + rlimits)"},
  "worker_processes": true
}

On a host that cannot create the namespaces the same block is {"state": "unavailable", "reason": "..."} and strict refuses. Design rule (blueprint §9): a cooperative in-process check is not a sandbox. See ADR-0021 for the Linux construction, L1 scope and non-goals; ADR-0019 still governs every other platform.

1.1 @exec is not a grant

The @exec heading mode introduced by RFC-LANG-0002 is a compile-time projection, not an authorization. A payload in @exec still passes the same CapabilityHost checks, budgets and sandbox rules; a payload in @knowledge produces no IR and therefore cannot cause a host effect at all. No heading annotation, keyword or Markdown container can grant a file/network/process capability, and permission flags never change the AST→IR result.

1.2 Strict failure and --json contract

Strict execution is a two-phase operation: detect, then launch. Availability observed by inspect --platform is a snapshot, not a permanent promise. The runtime therefore re-probes before launching; a failure after a successful probe is a refusal, never a fallback.

--json consumers must receive exactly one JSON document on stdout for every outcome:

  • a verified child in --json mode produces that document itself and its bytes and exit code are forwarded unchanged;
  • a control-plane failure (probe, supervisor spawn, namespace construction, timeout, capture) produces one CLI error document {"command": ..., "success": false, "error": ..., "code": ...};
  • output that is empty or is not exactly one JSON document is rejected as strict.child_invalid; raw supervisor text is never forwarded as program stdout.

The stable codes are strict.probe_unavailable, strict.spawn_failed, strict.namespace_failure, strict.timeout, strict.resource_limit, strict.child_invalid, strict.identity_invalid and strict.unsupported_grant (ADR-0053). A marker alone never authorizes strict execution: Linux verifies the namespace/privilege identity, and every other platform rejects MARQDO_STRICT_ACTIVE outright (ADR-0053).

2. Capability grants

Capabilities (crates/marqdo-host/src/capability.rs) is the effective grant set for one run:

Field Default (pure/project) Used by
console: bool true resource.open("stdio","stdin",["read"]), resource.open("stdio","stdout"/"stderr",["write"]) and the stdlib console chain; a missing grant is permission.denied and stdout/stderr are also bounded by the output budget
math: bool true math.* host ops
fs_read: Vec<String> empty fs.read and project resource.open("file", path, ["read"])
fs_write: Vec<String> empty fs.write and project resource.open("file", path, ["write"])
net_hosts: Vec<String> empty exact host/IP grants for `resource.open("tcp"
env: Vec<String> empty exact-name grants for resource.request("env.get", [name]); no glob and no env.list; --allow-env <NAME> adds a grant (ADR-0043)
process_programs: Vec<String> empty exact program paths for resource.request("process.run", [program, args]); no shell, env_clear, project root cwd, 5 s timeout, 1 MiB stdout/stderr cap; --allow-process <PROGRAM> adds a grant (ADR-0044); not an OS sandbox
random: bool false OS entropy for resource.request("random.bytes", [count]); --allow-random enables it; project mode only (ADR-0046)
process: bool false required together with foreign for foreign.run/foreign.call
foreign: bool false foreign kernels
plugins: Vec<String> empty plugin.invoke and StdHost::load_plugin
resource_providers: Vec<String> empty registered in-process/worker/C-ABI ResourceProvider kinds; project mode only; helper processes and native libraries are trusted, not sandboxed (ADR-0033/0035/0036); local manifests are host-installed, unsigned (ADR-0037); the C worker helper is a trusted process boundary, not a sandbox (ADR-0038)

Requests versus grants:

  • marqdo.yaml capabilities: is a request. It is parsed and validated (crates/marqdo-package/src/lib.rs) but is not consulted when building the host capability set: only CLI flags/grant APIs produce grants. Declaring a capability never grants it.
  • marqdo.yaml plugins: likewise declares requirements only; it does not install, load or authorize a plugin.
  • CLI grant flags (crates/cli/src/main.rs::SandboxArgs):
    • --allow-read <GLOB> and --allow-write <GLOB> (repeatable, project mode only);
    • --allow-net <HOST> (repeatable, exact host name or IP literal, project mode only, ADR-0034/0047);
    • --allow-env <NAME> (repeatable, exact environment variable name, project mode only, ADR-0043);
    • --allow-process <PROGRAM> (repeatable, exact program path, project mode only, ADR-0044; the child is not sandboxed);
    • --allow-random (project mode only; OS entropy, ADR-0046);
    • --allow-foreign (project mode only; sets foreign = true and process = true);
    • --allow-plugin <PACKAGE-NAME> (repeatable, project mode only; matches the manifest plugin.name, not the call namespace);
    • --plugin <PATH> (repeatable, loads a plugin after the trust check);
    • --trust-native-plugin (explicit in-process native trust decision).
  • There is no plugin-worker flag in this build. Network and environment grants are exact and project-mode only.

2.1 Filesystem grants are project-relative globs

fs.read/fs.write paths go through ProjectFs::resolve before the capability check:

  • empty paths, absolute paths, .. components and root/prefix components are rejected with permission.denied;
  • . components are dropped; the normalized relative string uses /;
  • the capability patterns are matched with the glob crate against that normalized relative path; a leading ./ is stripped on both sides, so --allow-read ./data/** matches data/a.csv and does not match secrets/key; --allow-write output/** does not grant reads;
  • an invalid glob pattern is treated as non-matching (Pattern::new(...). unwrap_or(false)), so it silently grants nothing (there is no diagnostic).

Additional FS limits (crates/marqdo-host/src/fs.rs):

  • reads are capped at MAX_READ_BYTES = 8 MiB, must be regular files, must be UTF-8, and the canonicalized path must stay inside the project root;
  • writes create parent directories; before writing, the closest existing ancestor is canonicalized and must stay inside the project root.

Known caveat: for writes, only the closest existing ancestor is checked and the final write uses the non-canonical path. A symlink created between the check and the write (TOCTOU) could escape the root. Reads canonicalize the final path before reading. Full symlink/race hardening is an open P4 item.

2.2 Checks happen at execution points, not only at parse time

Host::invoke is the only path to effects. StdHost::dispatch routes every HostOp and the checks run immediately before the effect:

  • fs.read → capabilities.check_read(path) then ProjectFs::read;
  • resource.request("fs.*") → check_read/check_write before path resolution;
  • resource.request("env.get") → capabilities.check_env(name) before std::env::var;
  • resource.request("process.run") → capabilities.check_process(program) before Command::spawn;
  • resource.request("random.bytes") → capabilities.check_random before getrandom::fill;
  • fs.write → check_write then ProjectFs::write;
  • math ops → capabilities.math (in crates/marqdo-host/src/kernels.rs);
  • foreign ops → capabilities.check_foreign (requires foreign && process);
  • plugin invoke → the complete export name must be registered, then capabilities.check_plugin(package) on every call; loading also checks the same grant once;
  • plugin host-capability requests → CapabilityBroker runs a fixed chain: decide (package, effective grants, trusted vendor policy, payload shape) → a trusted CapabilityHandler must be installed for the exact capability name → a package-bound handler must own the trusted package identity → the handler executes under the broker budget and reply limit → one audit entry. Granted fs.read executes through FsReadHandler, which resolves the path with ProjectFs and re-checks the read grant on the normalized project-relative path before returning the real file bytes. Every capability without an installed handler — other core capabilities (fs.write, net.request, process, console/math/log/clock/random/artifact.record) and vendor capabilities without a trusted executor — returns capability.unavailable; it never pretends an empty reply is success. A vendor name with no trusted policy entry is permission.denied (ADR-0011, ADR-0057);
  • print → output budget only.

Static analysis may warn, but it never replaces these runtime checks. There is no host bypass for tests: tests use the same StdHost/Capabilities types.

2.3 Executable capabilities need an installed trusted handler

CapabilityBroker (crates/marqdo-host/src/capability.rs, ADR-0057) separates being authorized from being executable:

Question Registry Failure
Is this capability granted for this trusted package? Capabilities / vendor_policy permission.denied
Is a trusted executor installed for this exact name? handlers: BTreeMap<String, HandlerEntry> capability.unavailable
Does a package-bound handler own the claimed package? HandlerEntry.package permission.denied

Execution order is fixed: decide → handler installed → package binding → handler execution under the broker Budget and reply limit → audit. A name in vendor_policy is policy only; it does not install anything, and it cannot turn into an executor. A core capability granted by Capabilities but missing a handler is capability.unavailable, never an empty Ok.

  • Registration is host-only. register_handler(handler) / register_package_handler(package, handler) are called by the trusted host / embedder composition root; no manifest, plugin or wire message can reach the registry. The registry key is handler.name(), so a handler cannot be registered under a name it does not declare.
  • Rejected names (stable codes): empty/whitespace/malformed names are capability.handler_invalid; names under the reserved host prefixes marqdo. / host. are capability.handler_reserved; an already registered name (including the built-in fs.read) is capability.handler_conflict. A rejected handler is not installed.
  • fs.read is the first migrated standard handler (FsReadHandler), installed by CapabilityBroker::with_fs(fs). It keeps the ProjectFs root, traversal/absolute-path rejection, regular-file/UTF-8/8 MiB limits and the glob grant, and it re-checks the grant against the normalized path at the execution point. The broker reply limit (MAX_CAPABILITY_REPLY_BYTES by default) bounds the returned bytes; an oversize reply is resource.exhausted.
  • Audit: every decision records package, capability, granted/denied/unavailable, a reason and duration_micros. Payloads and file contents are never recorded, so ordinary secrets do not appear in the audit log.
  • Budgets: handlers receive &mut Budget and can charge fuel/output; the broker also enforces the reply byte limit. Native handlers still run in-process and are trusted code, not a sandbox.

Known gaps in this build: the broker trace_id is settable (with_trace_id) but host.rs does not supply it yet; and a broker-level resource.exhausted reply keeps its stable code but the host plugin-error mapping currently collapses PluginError kind capability to the permission.denied runtime category (changing that mapping is a follow-up).

3. Resource budgets

Budget (crates/marqdo-host/src/budget.rs) is per run:

Limit Default Enforcement
evaluation fuel DEFAULT_FUEL = 5_000_000 1 unit per statement, 1 per expression node, 1 per branch condition, 1 per loop iteration; exhaustion → resource.exhausted
call stack depth DEFAULT_STACK_LIMIT = 256 checked before every function call → resource.exhausted
program output bytes DEFAULT_OUTPUT_BYTES = 1 MiB print, stdio/stdout and stdio/stderr output plus fs.read contents are charged; overflow → resource.exhausted
wall clock none by default --timeout <seconds> sets a deadline; checked between statements/iterations and before loop conditions → resource.exhausted. It does not preempt a running host call.
addressable memory 64 MiB committed total, 16 MiB per region, 64 KiB pages; --max-memory / --max-region-memory override Runtime-owned MemoryStore (ADR-0022); create/grow checks both budgets before allocation and failed growth is atomic → resource.exhausted. Memory is pure computation and needs no capability grant in any mode.
resource streams stdin reads over a shared run cursor; stdout/stderr writes charged to the existing output budget; file reads and writer buffers bounded by the 8 MiB per-file limit Host-owned resource table (ADR-0025/0026/0040); output budget is checked before a stdout/stderr write is buffered; a failed check produces no bytes; writer buffers flush only on close. No separate I/O byte budget yet.

CLI overrides (run/call): --fuel, --stack, --max-output, --timeout. Additional fixed limits:

  • foreign kernel: StdHost.foreign_timeout = 10s and foreign_output_limit = 1 MiB (stdout+stderr combined) per call. These are not configurable from the CLI in this build;
  • plugin wire: request ≤ 16 MiB, result ≤ 64 MiB, error ≤ 1 MiB, metadata ≤ 64 KiB (see include/marqdo_plugin_abi_v1.h);
  • math kernel: 1 MB source, 256 parse depth, 512 AST nodes, exponent ≤ 10,000, exact-result ≤ 1,000,000 bits (crates/marqdo-math).

These are cooperative budgets in the interpreter and the supervisor. Only foreign child processes are terminated by the host (kill on timeout/output overflow). In-process math and native plugins are not OS-limited.

4. Effect tracing

EffectTrace records one entry per Host::invoke when enabled (--trace, or implicitly --json for run/call). Entry fields:

{"op":"print","module":"hello.mq.md","span":{"start":43,"end":57},
 "duration_micros":3,"result":"ok"}
  • op is the stable HostOp::name() (print, math.eval, foreign.call, fs.write, plugin.invoke, ...);
  • module is the initiating module id; span is the call site when known;
  • result is "ok" or the failure code (for example permission.denied);
  • argument values and file contents are not recorded, so ordinary secrets do not leak into traces by design. Paths are also not recorded today.
  • In text mode trace entries are printed to stderr (trace: ...); in JSON mode they are a trace array next to separate program stdout/stderr strings. Program output is always separate from diagnostics/trace, and stdio/stderr never shares the stdout buffer.

5. Foreign kernels: process isolation, not a sandbox

crates/marqdo-foreign/src/lib.rs runs Python/Node one-shot:

  • the program is written to a private temp file (mode 0600 on Unix), the child runs with a fresh private temp directory (mode 0700) as cwd, and both are removed on drop;
  • direct argv exec (no shell, no interpolation); stdin carries the versioned request envelope; stdout/stderr are always piped and captured;
  • the environment is cleared except PATH;
  • on Unix the child is a process-group leader (setpgid) and the wall-clock deadline/output overflow kills the whole group, including grandchildren; a 2-second reader grace prevents a surviving pipe holder from hanging the host. Non-Unix targets only kill the direct child;
  • a non-zero exit is data (RunStatus::Failed, stderr preserved); a timeout is TimedOut for run and an error for the typed call.

What this is not: an OS sandbox. The child runs with the host user's privileges, can open absolute paths, use the network and spawn further processes. Clearing the environment and changing cwd are hygiene, not confinement. Grandchildren are not killed when the direct child is killed. Strict L1 rejects Foreign execution before any sandbox starts, so foreign execution is only reachable in project mode with --allow-foreign. L2 inheritance is future work (ADR-0021).

6. Plugins: in-process native trust boundary

  • The only plugin transport implemented in this build is in-process native loading (libloading) after manifest.validate(), exact target-triple artifact resolution, sha256 verification, ABI negotiation and export metadata cross-checking (crates/marqdo-plugin-abi/src/loader.rs).
  • Loading requires three independent things:
    1. --plugin <manifest-or-dir> names the plugin;
    2. --trust-native-plugin is an explicit unsafe trust decision;
    3. the manifest package name is granted with --allow-plugin <name> (checked again on every invocation). Without trust, StdHost::load_plugin returns permission.denied with the hint that no isolated worker transport exists. There is no isolated worker, Wasm component or out-of-process plugin path in this build.
  • A native plugin runs in the host address space. It can read host memory, call arbitrary syscalls, bypass the capability model entirely and crash or corrupt the host process. A C ABI is not a sandbox. --trust-native-plugin must be treated as equivalent to running the plugin's code as the host user.
  • Every export is registered by its complete dotted name; a plugin may export several prefixes, and a conflicting name rejects the whole load atomically. check --plugin <manifest> (or run/call --plugin) validates export signatures statically without loading a shared library.
  • Plugin call sites are written as <export>(...) in a function body (for example solver.add(a=20, b=22)). The namespace is derived from the first manifest export prefix. Calls are made through HostOp::PluginInvoke with a typed wire_version 1 request; the plugin's typed response becomes a Marqdo value (Unit when absent). Named arguments are required in this build; positional plugin arguments are runtime.type.
  • Plugin errors map to structured runtime errors: ABI → plugin.abi_mismatch, capability → permission.denied, wire/value → runtime.type, unsupported → kernel.unavailable, everything else → plugin.failure. An unloaded namespace is symbol.unknown.
  • The plugin's host-capability requests are denied by default: the CLI loads plugins with LoadOptions::default(), i.e. no capability handler, so every request_capability call fails. The requires: list in plugin.yaml is only a declaration.
  • marqdo plugin inspect never loads code: it validates the manifest and verifies the artifact hash only.
  • Installation, inspection, loading and authorization are four separate decisions. Copying a plugin into the project or declaring it in marqdo.yaml grants nothing.

Known limitations: single namespace per plugin (derived from the first export), no hot unload/reload, no plugin capability bridge from this CLI, native crash is a host crash, plugin.invoke cannot be used in pure/strict mode.

7. check and inspect are side-effect free

Neither command executes project code. Specifically:

  • check parses, indexes, maps to IR, runs static call validation and reports diagnostics. It never calls a function, never spawns a foreign kernel, never loads a native plugin and never writes files.
  • inspect with --ast, --units, --signature, --ir, --refs or --platform does the same static work and prints metadata. --refs may read other project source files on demand to resolve link targets, but only parses/analyzes them; nothing is executed or written.
  • plugin inspect reads plugin.yaml, validates it, resolves the artifact for the host target and computes/compares its sha256. It does not dlopen, instantiate or call the plugin ("loaded": false in JSON output).
  • run/call first perform the same static load; if any diagnostic has error severity, execution is refused before the host or any kernel is touched.
  • fmt only normalizes line endings/trailing newlines and writes the document when not in --check mode; it does not execute anything.

8. Open security items

  1. Strict L1 is implemented and VM-verified on Linux (ADR-0021); strict L2 (Worker/Foreign inheritance) is not implemented, so those surfaces are refused in strict.
  2. console capability is not enforced; print is always allowed.
  3. Network capability exists in the model but no host operation uses it; there is no --allow-net flag.
  4. Write-path symlink TOCTOU (see §2.1).
  5. Plugin worker crash isolation and the worker capability bridge are implemented (not usable in strict L1): both native and worker plugins can reach the broker's fs.read handler when granted (ADR-0057). Only fs.read has a registered handler; every other core capability and every vendor capability without an explicitly registered trusted handler remains capability.unavailable, and there is no API to register one from a plugin, manifest or CLI flag. Broker execution is a cooperative API contract, not an OS sandbox; a worker and a native plugin can still issue direct syscalls. Follow-ups: wire the run trace_id into the broker audit and teach the host plugin-error mapping to report resource.exhausted as a resource category rather than permission.denied.
  6. plugin_handle wire values are structurally reconstructable, but host authority comes only from the PluginHandleTable owner/run/generation check; untrusted plugin/foreign output containing a handle is rejected (ADR-0014, value-wire.md §4.1).
  7. Budgets do not cover all allocation paths (for example list/table growth inside the interpreter is bounded only by fuel and memory).
  8. --json run output includes the full program stdout string; that is the program's own output, not a host secret, but it is embedded in the JSON document consumers must parse.