Files
pmg/config/config.template.yml
T
c47776db27 feat(proxy): persistent proxy server mode (#351)
* refactor(flows): extract SetupCACertificate for reuse

Move the CA load/generate/merge logic out of proxyFlow into an exported
flows.SetupCACertificate so the persistent proxy server can reuse it.

* feat(proxy): add persistent proxy server with start/stop/env/status

Introduces 'pmg proxy' commands backed by internal/proxyserver: a long-lived
MITM proxy that intercepts package managers via env vars (no shims). Supports
--daemon (Unix), --state, --port; generic 'env' output that skips cert vars
when the CA is OS-trusted; opt-in 'stop --fail-on-violation' (fail-closed on
crash) with a synchronous cloud event flush; and the malysis analysis cache.

* feat(action): add server-mode for persistent proxy

When server-mode=true the action starts the proxy daemon and injects proxy
env vars into the job instead of installing shims.

* test(proxy): add persistent proxy server E2E workflow

* docs(readme): document persistent proxy server mode

* fix(proxy): create cache dir before writing state file and daemon log

On a fresh CI runner the cache directory does not exist yet; os.OpenFile and
os.WriteFile do not create parent dirs, so 'pmg proxy start --daemon' failed
with 'no such file or directory'. MkdirAll the parent before writing.

* docs: add persistent proxy server architecture doc

* refactor proxyserver

* fix(proxy): always emit cert env vars instead of skipping on OS-trust status

npm/pip/yarn/requests trust the MITM CA inconsistently across tools, versions,
and configs; many still use bundled CA stores. Always emitting the cert-path
env vars is the conservative choice that works regardless, and is harmless for
tools that read the OS store (they ignore the vars). Skipping them when a
system CA exists would silently break any tool still on a bundled store.

* refactor(proxy): drop redundant audit init in daemon; rely on main.go

main.go's PersistentPreRun already initializes the audit pipeline for every
command (including the daemon's re-exec'd child) and closes it at process exit.
Re-initializing in proxyserver.Run created a second auditor and a second
cloud-sync WAL connection, orphaning the first. Removing it makes the daemon
consistent with the normal proxy flow, which never self-initializes audit.

* fix(proxy): bypass proxy env when flushing events to cloud on stop

pmg proxy stop inherits HTTP(S)_PROXY (injected by 'pmg proxy env') pointing at
the PMG proxy it just shut down. The cloud sync gRPC client honored those vars
and routed api.safedep.io through the dead proxy, failing with 'connection
refused' so no events were delivered. Clear the proxy env vars before the sync
so PMG's own cloud traffic goes direct.

* chore(proxy): address review feedback

- configurable bind host via proxy.server.listen_host (default loopback)
- proxy commands use ui.ErrorExit instead of returning errors to cobra
- rename errcode to ProxyPolicyViolation (covers malware + cooldown)
- share cloud sync via audit.DrainToCloud (de-dup with cmd/cloud/sync)
- centralize proxy CA bundle path in certmanager
- docs: persistent proxy cert trust + bind address

* fix(proxy): show real message on fail-on-violation error

stopExitError set only WithMsg, but ui.ErrorExit renders HumanError, so the
framed error showed 'no human-readable message available'. Set both from one
string, and emit the framed error before the stdout summary so the blocked
count is stated once.

* fix(proxy): flush cloud events from the daemon, not stop

The stop process inherits HTTP_PROXY (from 'pmg proxy env'), so its cloud
client routed api.safedep.io through the already-stopped proxy and failed with
connection refused. Move the flush into the daemon's shutdown, which has no
proxy env (it started before env injection) and dials SafeDep directly.

- daemon flushes on shutdown via audit.DrainToCloud and records the result in
  the state file; stop surfaces it (on both success and fail-on-violation
  paths) since the daemon's own logs aren't visible to stop
- coordinate stop's wait with the daemon shutdown budget; on timeout, error
  out without reading stale state or deleting the file (fail-closed)
- persist blocked count before the flush so the gate stays correct if the
  flush hangs or the daemon is killed mid-flush
- remove now-redundant cloud_flush.go

* disable auto-sync for proxy cmds

* feat(proxy): periodic cloud sync + move proxy env vars to packagemanager

- daemon runs a periodic cloud-sync ticker so the shutdown flush stays small;
  the run total is reported by stop, and shutdown timeouts are coordinated
- move EnvVarForProxy from config to packagemanager (it is package-manager
  knowledge); the shared function now builds the proxy URL and NO_PROXY itself,
  removing the duplicated construction in the per-command and persistent paths
- relocate the #319 yarn and #339 IPv6 regression tests alongside the function
- enable cloud sync in the persistent-proxy E2E workflow and fix the stale
  internal/proxystate path filter

* refactor(proxy): rename cloudFlushLockTimeout to cloudFlushLockWait

Consistent timeout naming: *LockWait is the lock-acquire bound, *Timeout is
the sync-RPC bound. Previously the final-flush pair was cloudFlushLockTimeout
vs cloudFlushTimeout — two lookalike names for different operations.

* refactor(proxy): extract cloudFlush and trim duplicate shutdown comments

The shutdown's final-flush block is now a cloudFlush helper, symmetric with
startCloudSyncLoop (one-shot vs loop). Removed the triplicated ticker/lock
contention comments, keeping the contract on the function doc and one-line
pointers at the call sites.

* docs: update persistent proxy cloud sync to daemon-owned model

The daemon now owns cloud delivery (periodic sync while serving + final flush
on shutdown); stop signals it, waits, and reports the result. Rewrite the Cloud
event sync section, fix stop attributions, add the cloud_sync state field, and
update the sequence diagram.

* docs: move Usage section up below How it works

Put the copy-paste recipes near the top so users find them before the internals.

* refactor(proxy): address PR review feedback

- configurable bind host/port via --host/--port flags + config (listen_host,
  listen_port), bound directly to config fields per PMG's flag pattern
- daemon log path via --log-file and readiness timeout in ProxyDaemonConfig;
  Daemonize no longer owns path policy (caller validates, fails fast)
- gate periodic cloud sync on auto_sync; suppress detached background sync for
  proxy commands instead of flipping the flag
- pmg proxy env --export emits shell-quoted lines for eval (spaces survive)
- extract shared flows.BuildCachedMalysisAnalyzer, dropping the analyzer+cache
  duplication between proxy flow and proxy server
- add internal/proxyserver/doc.go documenting the package + boundary vs flows
- E2E: assert malicious installs are blocked (drop continue-on-error)
- docs: trim Commands/State-file to user contracts; refresh bind address

* refactor(proxy): proactive alignment fixes from whole-PR review

- gate the shutdown cloud flush on auto_sync too, matching the periodic ticker
  (auto_sync consistently controls all daemon-driven cloud delivery)
- ResolveStatePath takes cacheDir instead of *RuntimeConfig, keeping state.go
  free of config dependency
- drop the empty-host comment in listenAddr; keep the loopback guard so a blank
  host never silently binds all interfaces

* fix: Decouple localdb with malysis analyser construction

* fix: Persist global args before proxy server daemon exec

* fix: GitHub Action for cloud auto-sync in server mode

---------

Co-authored-by: Abhisek Datta <abhisek.datta@gmail.com>
2026-06-26 11:19:28 +05:30

250 lines
9.9 KiB
YAML

# PMG configuration template. Customize this file as needed.
# https://github.com/safedep/pmg
# Enable transitive dependency resolution. Default is true.
transitive: true
# Maximum depth of transitive dependencies to resolve. Default is 5.
transitive_depth: 5
# Include dev dependencies in the dependency graph. Default is false.
include_dev_dependencies: false
# UI verbosity level. Valid values: silent, normal, verbose. Default is normal.
# silent: PMG is hidden from the user except for errors and malicious package detection
# normal: Show minimal status updates
# verbose: Show verbose status updates and detailed information
verbosity: normal
# Enable paranoid mode. In paranoid mode, PMG will treat suspicious packages
# as malicious packages
paranoid: false
# Disable anonymous telemetry. Default is false.
disable_telemetry: false
# Skip event logging. Default is false.
# When skip_event_logging is false, all events will be logged to file. These events are useful for audit
# trail and incident response on systems using PMG. Set this config to true to disable event logging.
skip_event_logging: false
# Event log retention days. Default is 7.
# This is the number of days to retain event logs.
event_log_retention_days: 7
# Proxy configuration.
# When enabled, PMG uses a proxy-based interception approach instead of the
# default guard-based analysis. The proxy intercepts package manager requests in real-time
# and analyzes packages as they are downloaded. Proxy mode may not work in all environments,
# and can be disabled to fall back to the guard-based analysis.
proxy:
enabled: true
# When true, only install commands are proxied. Other commands
# (e.g., npm ls, pip list) bypass the proxy and execute directly.
install_only: false
# Per-package-manager commands to skip proxying (only applies when install_only is true).
# Example:
# skip_commands:
# pip: ["list", "show"]
skip_commands:
npm: []
# Persistent proxy server (`pmg proxy start`) settings.
server:
# Host the persistent proxy binds to. Defaults to 127.0.0.1 (loopback),
# which keeps the MITM proxy private to the host (the right choice for CI
# and local use). Set to 0.0.0.0 or a specific interface ONLY for a
# deliberately hosted deployment: a non-loopback bind exposes the proxy,
# and every client must trust the PMG CA. The --host flag overrides this.
listen_host: 127.0.0.1
# Port the persistent proxy binds to. 0 means a random free port. The
# --port flag overrides this.
listen_port: 0
# Trusted packages are packages that are trusted by the user and will be ignored by the security guardrails.
# This is useful for packages that are known to be safe and are used in the application.
# Example:
# - purl: pkg:npm/@safedep/pmg
# reason: "PMG is a trusted package for PMG"
#
# When a package is specified without an explicit version, all versions of the package will be trusted.
# This feature should be used with caution and should be used for minimal set of packages.
#
# When a package is specified with an explicit version, only that version will be trusted.
#
# The purl is the package identifier and the reason is the reason for trusting the package.
# PURL specification: https://github.com/package-url/purl-spec
trusted_packages:
- purl: pkg:npm/@safedep/pmg
reason: "PMG is a trusted package for PMG"
# Sandbox configuration (EXPERIMENTAL)
# When enabled, package managers run in sandbox environments with restricted
# filesystem, network, and process execution access. This provides defense-in-depth
# protection against malicious install scripts and supply chain attacks.
#
# When sandboxing is enabled for a package manager, policy violations will block execution
# (no "warn-only" mode). Sandboxing itself can be disabled globally or per package manager.
#
# Currently supported platforms:
# - macOS (using Seatbelt sandbox-exec)
# - Linux (using Bubblewrap with namespace isolation)
# - Windows (planned)
#
# Platform-specific limitations:
# - Linux: Filesystem permissions use coarse-grained bind mounts. Glob patterns (e.g., *.txt)
# are expanded at policy translation time, but entire directories may be mounted rather than
# individual matching files. This is less precise than macOS regex-based filtering.
# - macOS: Network filtering is limited (all-or-nothing for most policies).
sandbox:
# Enable sandbox mode (opt-in, default: false for backward compatibility)
enabled: false
# Controls scope of sandbox enforcement:
# - true: sandbox applies to all package manager commands
# - false (default): sandbox only applies to install commands, others run unrestricted
# Requires 'enabled: true' and per-PM policies to be active. May break workflows expecting unrestricted commands.
enforce_always: false
# Policy templates define policy profiles by name and path.
# They can be used to override a built-in profile or create a custom profile.
# Note: Custom profiles loaded via policy_templates can inherit from built-in
# profiles using the 'inherits' field in the YAML (e.g., inherits: npm-restrictive).
# Inheritance allows you to extend a base profile with additional permissions.
policy_templates:
# Name for the template. Can be used to override a built-in profile or create a custom profile.
# Path is the path to the template file.
# Relative path can be used to reference a template file in the config directory (example: ./npm-restrictive.yml)
npm-restrictive-override:
path: ./profiles/npm-restrictive.yml
# Per-package-manager sandbox policies
# Each package manager can have its own policy to account for unique security characteristics
policies:
# npm ecosystem. npm-restrictive is the shared base profile; each package
# manager maps to a leaf profile that re-allows only its own environment
# variables.
npm:
enabled: true
profile: npm # Built-in profile, template name, or path to custom YAML
pnpm:
enabled: true
profile: pnpm
npx:
enabled: true
profile: npx
pnpx:
enabled: true
profile: npx
yarn:
enabled: true
profile: yarn
bun:
enabled: true
profile: bun
# PyPI ecosystem. pypi-restrictive is the shared base profile; each
# package manager maps to a leaf profile that re-allows only its own
# environment variables.
pip:
enabled: true
profile: pip
pip3:
enabled: true
profile: pip
pipx:
enabled: true
profile: pipx
poetry:
enabled: true
profile: poetry
uv:
enabled: true
profile: uv
# Dependency cooldown blocks installation of package versions published within
# a configurable time window.
dependency_cooldown:
enabled: true
days: 5
# Per-control skip list of packages exempt from the cooldown window.
# Packages here are STILL malware-scanned — only the cooldown wait is waived.
# Use it for first-party / internal packages that must be installed immediately
# on release (e.g. to sanity-test a freshly published version).
#
# To bypass every PMG control (malware analysis, cooldown, and any future
# controls) for a package, add it to the top-level trusted_packages list
# above instead. Trusted packages are automatically cooldown-exempt; you do
# not need to repeat them here.
#
# A PURL without a version skips cooldown for ALL versions of the package; a
# PURL with a version skips cooldown for that version only. Example:
# skip:
# - purl: pkg:npm/my-internal-sdk # all versions
# reason: "First-party SDK; sanity-tested immediately on release"
# - purl: pkg:npm/another-internal-pkg@1.2.3 # only 1.2.3
# reason: "Pin a specific just-published build"
skip: []
# Persistent analysis cache (opt-in). Caching is analyzer-specific, so config is
# nested per analyzer; today only the Malysis (malware) analyzer has a cache.
#
# By default PMG re-screens the whole resolved graph against the analysis
# backend on every install. When enabled, clean (ALLOW) verdicts are cached on
# disk and reused across runs, so repeat installs of an unchanged graph are
# fast. Only ALLOW verdicts are cached — suspicious, malicious, and
# tenant-excluded verdicts are always re-evaluated.
#
# Trade-off: a version that was clean when first screened but is later flagged
# malicious is served from cache until its entry expires; `ttl` bounds that
# window. Keep `ttl` short if you prefer freshness over speed.
analysis_cache:
# Malysis is SafeDep's threat intelligence feed.
# Set cache TTL to be lower than dependency cooldown period to have a compensating
# control in place for newly published packages whose verdict is cached.
malysis:
enabled: false
ttl: 24h
# Cloud sync configuration.
# When enabled, PMG audit events are synced to SafeDep Cloud for centralized visibility.
# Requires SAFEDEP_API_KEY and SAFEDEP_TENANT_ID environment variables for authentication.
cloud:
enabled: false
# Endpoint ID is not required. By default, it falls back to the machine's hostname.
# Set it only if you want to explicitly override the identifier for this endpoint.
endpoint_id: ""
# Auto sync drains the local cloud-sync WAL to SafeDep Cloud opportunistically
# at the end of each PMG invocation, gated by a per-host cooldown so it does
# not fire on every command. The drain runs as a short-lived detached child
# so the user-facing CLI returns immediately.
#
# Disable this in ephemeral environments (CI runners, throwaway VMs) where
# the detached child may be torn down before it finishes draining; in those
# environments, prefer an explicit `pmg cloud sync` at job-end.
auto_sync:
enabled: true
# Minimum gap between sync attempts. The lastrun timestamp is updated on
# every attempt (success or failure), so a failing cloud endpoint will not
# cause every PMG invocation to retry. Reduce this if you want a more
# aggressive retry cadence.
min_interval: 15m
# Hard timeout applied to a single background sync attempt.
timeout: 5m