Agent Scan CLI reference

August 28, 2026 · View on GitHub

This reference covers both the 2025-09-02 API used by Agent Scan v0.5.x and the risk-based 2026-07-10 API used by Agent Scan v0.6 and later. Commands and flags are documented once when their behavior is shared; version-specific behavior is called out where it differs.

Version selection

Agent Scan v0.5.x

Important

Agent Scan v0.5.x is planned for deprecation but remains supported for now. Pin a concrete v0.5.x release so an existing workflow does not silently switch APIs or output formats.

uvx snyk-agent-scan@0.5.17

v0.5.x uses issue codes and severity labels. Its JSON schema and CI ignore flag differ from v0.6 and later.

Agent Scan v0.6 and later

uvx snyk-agent-scan@latest

v0.6 and later uses named, scored risk indicators and separates risk ignores from operational failure-code ignores.

Agent Scan ships in two forms:

Entrypoint	Command	Notes
Standalone CLI	`snyk-agent-scan` or `uvx snyk-agent-scan@VERSION`	Python package; primary surface documented below
Snyk CLI extension	`snyk agent-scan --experimental`	Go wrapper that downloads a pinned binary and forwards flags; see Snyk CLI extension flags

Unless noted, flags apply to the standalone CLI. When no subcommand is given, scan is assumed.

Experimental output. Issue/risk names, JSON field names, and human-readable output may change between releases. See the project README for the stability notice.

Commands

Command	Description
`scan`	Discover and inspect MCP configs, agents, and skills, then request security analysis (default)
`inspect`	List tools, prompts, resources, and skills without requesting security analysis
`help`	Print top-level help
`evo`	Interactive flow: mint a push key, scan, upload to Snyk Evo, and revoke the key
`guard`	Install, uninstall, or show the status of Agent Guard hooks

Positional arguments

The scan-like commands accept optional paths:

snyk-agent-scan [scan] [CONFIG_FILE ...]
snyk-agent-scan inspect [CONFIG_FILE ...]
snyk-agent-scan evo [CONFIG_FILE ...]

Argument	Applies to	Description
`CONFIG_FILE ...`	`scan`, `inspect`, `evo`	One or more MCP configuration files, `SKILL.md` files, or directories. If omitted, well-known agent configuration locations are discovered automatically.

Shared options (`scan`, `inspect`, and `evo`)

Output and logging

Flag	Type	Default	Description
`--json`	boolean	`false`	Emit JSON on stdout instead of rich text. Non-JSON status lines are suppressed. The schema depends on the command and CLI version; see JSON output.
`--verbose`	boolean	`false`	Enable debug logging to stderr.
`--print-errors`	boolean	`false`	Show error details and tracebacks in the human-readable report.
`--print-full-descriptions`	boolean	`false`	Show full tool and skill descriptions rather than truncating them.
`--show-analysis-results`	boolean	`false`	Force synchronous analysis so scan results are returned and displayed. Overrides the default push-key behavior, which submits the scan asynchronously and returns no local results. Implied by `--ci` and the `evo` command. See Push keys and results.

Discovery and scope

Flag	Type	Default	Description
`--no-skills`	boolean	off	Skip agent skill discovery and analysis. Skills are included by default.
`--skills`	boolean	on (implicit)	Deprecated/no-op. Accepted for backward compatibility but redundant because skills are enabled by default. Prefer omitting it; use `--no-skills` to opt out.
`--scan-all-users`	boolean	`false`	Scan all readable user home directories (and WSL profiles on Windows), rather than only the current user.

With --no-skills, explicit skill paths, skill directories, and automatically discovered skills are all skipped.

Analysis and upload

Flag	Type	Default	Description
`--storage-file FILE`	string	`~/.mcp-scan`	Path for local scan state and cached results.
`--analysis-url URL`	string	version-specific	Analysis API endpoint; see the defaults below. With `SNYK_TOKEN`, the CLI rewrites it to the `/cli/analysis-machine` variant automatically. In v0.6+, any supplied `version` query parameter is replaced with `2026-07-10` so the URL cannot select an incompatible wire contract.
`--verification-H HEADER`	repeatable	—	Extra HTTP header for the analysis request, in `Name: value` format. Repeat for multiple headers.
`--skip-ssl-verify`	boolean	`false`	Disable TLS certificate verification for analysis and upload calls.
`--mcp-oauth-tokens-path PATH`	string	—	JSON file containing MCP OAuth tokens (`TokenAndClientInfoList` schema) for OAuth-protected remote MCP servers.

The default analysis URL is the main version-dependent value:

CLI version	Default analysis URL
v0.5.x	`https://api.snyk.io/hidden/mcp-scan/analysis-machine?version=2025-09-02`
v0.6 and later	`https://api.snyk.io/hidden/mcp-scan/analysis-machine?version=2026-07-10`

The URL version must match the request and response models used by the installed Agent Scan version.

Push key and machine identifier (v0.6 and later)

v0.6 flag	Type	Default	Description
`--push-key KEY`	string	—	Push key used to authenticate with the analysis server. Replaces the soon-to-be-deprecated `--control-server-H "x-client-id: <push-key>"` pattern below and does not require a `--control-server` block.
`--machine-id ID`	string	—	Non-anonymous identifier for this machine (for example, email, hostname, or serial number). Replaces the soon-to-be-deprecated `--control-identifier` below.

If both a new flag and its soon-to-be-deprecated equivalent are given, the new flag wins. Using --control-server-H for the x-client-id trick, or --control-identifier on a single --control-server block, emits a one-time deprecation warning pointing at --push-key / --machine-id.

Control server (enterprise upload, on deprecation path)

Upload analyzed results to one or more control servers, typically Snyk Evo. This upload mechanism is on a deprecation path and will be replaced.

Flag	Type	Default	Description
`--control-server URL`	repeatable	—	Upload destination. Each block must include a matching `--control-identifier`. Repeat for multiple destinations.
`--control-server-H HEADER`	repeatable	—	Header for the current `--control-server` block, such as `x-client-id: <push-key>`. Repeat within a block for multiple headers.
`--control-identifier ID`	repeatable	—	Non-anonymous identifier for the current block, such as an email, hostname, or serial number. Required for every `--control-server`.
`--no-bootstrap`	boolean	`false`	See historical bootstrap behavior. In v0.5.14 and later it is an accepted compatibility no-op.

Options following a --control-server apply to that server until the next --control-server:

snyk-agent-scan scan \
  --control-server "https://api.snyk.io/hidden/mcp-scan/push?version=2025-08-28" \
  --control-server-H "x-client-id: <push-key>" \
  --control-identifier user@example.com \
  --control-server https://other.example/push \
  --control-server-H "Authorization: Bearer token2" \
  --control-identifier serial-123

Historical bootstrap behavior

Agent Scan v0.5.4 through v0.5.13 sent a startup bootstrap request to the first eligible control server before doing other work. This applied to every command that accepted --control-server: scan, inspect, and evo. Therefore, in those releases, inspect --control-server ... sent a bootstrap request even though inspect did not call the analysis API. --no-bootstrap disabled that request.

Only the first configured control server received bootstrap. The request was restricted to the Snyk-managed /mcp-scan/push endpoint, bootstrap failures did not abort the command, and the eventual scan result could still be uploaded to every configured control server.

The bootstrap implementation was removed in v0.5.14. In v0.5.14 and later—including v0.6—the --no-bootstrap flag remains accepted for command-line compatibility but does not change behavior.

Push keys and stdio MCP servers (`scan` only)

Enterprise uploads include a push key, either via --push-key (v0.6 and later) or the soon-to-be-deprecated x-client-id header in --control-server-H. Agent Scan treats this as an unattended run, so there is no terminal consent prompt.

`--dangerously-run-mcp-servers`	Consent prompts	Stdio MCP subprocesses
not set (default)	skipped	Not started; servers remain in the result as configured, while remote servers and skills are still inspected
set	skipped	Started for every stdio server in the scanned configurations

Use --dangerously-run-mcp-servers only when a trusted fleet or CI job must execute configured stdio MCP commands.

Push keys and results

When a scan authenticates with a push key and the tenant has asynchronous analysis enabled, Agent Scan submits the scan to the async endpoint and returns immediately without local results—results are processed and reported on the Snyk side. This keeps unattended and fleet scans fast and fire-and-forget.

Pass --show-analysis-results to force synchronous analysis instead, so the scan waits for the response and displays results locally (and in --json output). The synchronous path is also selected automatically for --ci runs and the evo command, so --show-analysis-results is only needed for a plain push-key scan.

CI mode

--ci makes scan-like commands return a non-zero status when relevant security findings or operational failures remain. It must be paired with --dangerously-run-mcp-servers in non-interactive use.

Agent Scan v0.5.x

Flag	Type	Default	Description
`--ci`	boolean	`false`	Exit with code `1` if analysis issues or runtime failures remain.
`--ignore-issues-codes CODES`	string	—	Comma-separated issue and failure codes to ignore for CI, such as `W001,W015,X001`. Only valid with `--ci`; ignored codes are also removed from JSON output in CI mode.

Code	Meaning
`0`	No remaining issues or failures after ignores
`1`	Issues or unignored runtime failures remain
`2`	Invalid command-line usage or flag combination

Agent Scan v0.6 and later

Flag	Type	Default	Description
`--ci`	boolean	`false`	Exit with code `1` if security risks or operational failures remain.
`--ignore-risks NAMES`	string	—	Comma-separated risk names to omit from output and CI evaluation, such as `dangerous_words,suspicious_download_url`. Only valid with `--ci`.
`--ignore-failure-codes CODES`	string	—	Comma-separated failure codes to omit from CI evaluation, such as `X001,X007`. Errors remain visible in output. Only valid with `--ci`.

Risk names and failure codes are case-sensitive. Unknown values produce a warning and are not applied. inspect supports --ignore-failure-codes because it can encounter operational failures but has no risks to ignore. evo supports neither ignore flag.

Code	Meaning
`0`	No remaining risks or operational failures after ignores
`1`	Risks or unignored operational failures remain
`2`	Invalid command-line usage or flag combination

MCP server options

These options apply to scan, inspect, and evo in both CLI versions.

Flag	Type	Default	Description
`--server-timeout SECONDS`	float	`10`	Timeout for MCP server connections, including stdio handshakes and remote servers.
`--suppress-mcpserver-io BOOL`	boolean	see below	Suppress stderr from stdio MCP servers. Stdout carries JSON-RPC and is never shown. Accepted values: `true`, `false`, `1`, `0`, `yes`, `no`, `y`, `n`, `t`, `f`.
`--dangerously-run-mcp-servers`	boolean	`false`	Skip per-server consent and start every configured stdio MCP server. Required with `--ci` in CI/CD. Use only in trusted environments.

Run type	Default for `--suppress-mcpserver-io`
Interactive (`inspect`, or `scan` without a push key)	`false`; stderr is streamed with a `[server-name]` prefix
Unattended (push-key scan, `evo`, and similar flows)	`true`; stderr is hidden

Handshake and consent matrix for stdio MCP servers:

Command	Push key	`--dangerously-run-mcp-servers`	Start servers	Consent prompt
`inspect`	—	no	yes	yes
`inspect`	—	yes	yes	no
`scan`	no	no	yes	yes
`scan`	no	yes	yes	no
`scan`	yes	no	no	no
`scan`	yes	yes	yes	no
`evo`	yes (automatic)	no	no	no

Scanning MCP configurations executes the commands defined in them. Review consent prompts carefully or run inside a sandbox. See the README Security Warning.

`scan`

scan runs the full pipeline: discover → inspect → redact → analyze → optional upload.

Flag	Type	Default	Description
`--checks-per-server NUM`	integer	`1`	No-op. Accepted for backward compatibility; does not change behavior.

The analysis result is the main version difference:

Agent Scan v0.5.x

Uses the 2025-09-02 analysis API.
Reports issue codes and severity labels.
scan --json prints the path-keyed ScanPathResult schema.
Uses --ignore-issues-codes for both security issues and operational failures in CI.

Agent Scan v0.6 and later

Uses the 2026-07-10 analysis API.
Reports named risk indicators with scores from 0 to 1000.
Human-readable output is compact by default: clean servers and skills show counts by entity or file type, while risky components list only the entities or files connected to a risk and summarize the remainder.
--show-full-discovery expands the human-readable report to list every MCP tool, prompt, resource, resource template, and skill file. It does not affect JSON output, which is always complete, or inspect, which always lists all locally inspected details.
scan --json prints a ScanResponse containing scan_path_responses.
Uses --ignore-risks and --ignore-failure-codes independently in CI.

v0.6 flag	Type	Default	Description
`--show-full-discovery`	boolean	`false`	Expand human-readable `scan` output to list every MCP entity and skill file instead of the compact summary.

Configuration file

Load argument values from a YAML file instead of passing them all on the command line. Applies to scan, inspect, and evo (not guard). Any flag from the sections above — Shared options, MCP server options, and scan options — is settable.

Flag	Type	Default	Description
`--config-file FILE`	string	—	Path to a YAML file supplying argument values. Errors (missing file, invalid YAML, non-mapping top level) exit with code 2.

Precedence. Values resolve as code defaults < config file < explicit CLI flags. A flag you also pass on the command line always wins over the file; a flag you omit takes the file's value, or the built-in default if the file doesn't set it.

Keys. Use the flag's long name with dashes or underscores — server-timeout and server_timeout are equivalent. Values are native YAML types (true/false, numbers, strings, lists), so --suppress-mcpserver-io becomes suppress_mcpserver_io: true and --dangerously-run-mcp-servers becomes dangerously_run_mcp_servers: true. Unknown keys are ignored with a warning on stderr.

Validation. Values are checked the same way argparse checks CLI input: the flag's type converter is applied, choices are enforced, and boolean/scalar/list shapes must match (a boolean flag rejects non-boolean values, a list option rejects a mapping, etc.). String forms of booleans are normalized (skip_ssl_verify: "false" → false). An invalid type, value, or shape exits with code 2.

Scalars vs. blocks/lists. Scalar options (server_timeout, analysis_url, booleans, …) override field-by-field. Block/list options — the positional files, repeatable headers like verification_H, and the control_servers block — use complete replacement: if you pass any CLI flag for that structure, the file's entire array for it is discarded (no element-wise merge) and rebuilt from the CLI alone.

Control servers. The --control-server / --control-server-H / --control-identifier block is expressed as a control_servers list; headers may be a mapping or a list of Name: value strings. Passing any one of those three flags on the CLI replaces the whole control_servers list from the file.

# agent-scan.yaml
json: true
skills: true              # set false to skip skills (equivalent to --no-skills)
scan_all_users: true

# MCP server options are settable too
server_timeout: 30
suppress_mcpserver_io: true
dangerously_run_mcp_servers: true

verification_H:
  - "X-Trace: abc123"
files:
  - ~/.cursor/mcp.json
  - ~/.claude/skills

# Push key and machine identifier (v0.6 and later)
push_key: <push-key>
machine_id: user@example.com

# Legacy control-server block (superseded by push_key/machine_id above)
control_servers:
  - url: https://api.snyk.io/hidden/mcp-scan/push?version=2025-08-28
    identifier: user@example.com
    headers:
      x-client-id: <push-key>

# Use the file as-is
snyk-agent-scan scan --config-file agent-scan.yaml

# ...but override one scalar for this run (CLI wins)
snyk-agent-scan scan --config-file agent-scan.yaml --server-timeout 5

# Passing --control-server discards the file's control_servers list entirely
snyk-agent-scan scan --config-file agent-scan.yaml \
  --control-server https://other.example/push --control-identifier host-1

Notes:

push_key / machine_id are ordinary scalars in the precedence cascade (code defaults < config file < explicit CLI flags), but they resolve against the soon-to-be-deprecated --control-server-H / --control-identifier flags rather than against each other's own CLI flag. Concretely: if you pass the legacy --control-server-H "x-client-id: ..." / --control-identifier on the command line but the config file sets push_key / machine_id, the file's value wins, because only an explicit --push-key / --machine-id on the command line — not the legacy flags — is checked as an override. Passing --push-key / --machine-id explicitly on the command line still overrides the file as usual.

skills is on by default. In YAML, set skills: false to disable skill scanning (equivalent to --no-skills); there is no separate no_skills toggle — the skills key is the single source of truth.

dangerously_run_mcp_servers: true in the file has the same effect as the flag, including satisfying the --ci requirement that it be set — a config file can therefore enable stdio MCP server execution. This is intended, but review config files with that in mind.

On/off flags set to true in a file cannot be turned off on the command line. Flags such as json, scan_all_users, and dangerously_run_mcp_servers are store_true and have no --no-… counterpart, so once a config file sets them to true there is no CLI flag to override them back to false for a single run — edit or omit the file instead. (skills is the exception: it has --no-skills.)

`inspect`

inspect uses the shared discovery and MCP options but does not request security analysis. It prints the locally discovered servers, skills, tools, prompts, resources, and resource templates. Its behavior is shared across the documented versions except for the historical bootstrap behavior in v0.5.4–v0.5.13.

`evo`

evo uses the scan discovery, inspection, analysis, and upload flow:

Prompt for a Snyk tenant ID and API token.
Mint a short-lived push key (x-client-id).
Run scan with the Snyk push endpoint configured automatically.
Revoke the push key when finished.

evo always authenticates with the push key it mints for that run. In v0.6 and later, an explicit --push-key on the command line is ignored — it never substitutes for the minted key.

--ci is accepted for compatibility but does not change evo output or exit status. Shared argument validation still requires it to be paired with --dangerously-run-mcp-servers. Use scan when CI risk/failure evaluation is required.

`guard`

Manage Agent Guard hooks for Claude Code, Cursor, and Codex:

snyk-agent-scan guard [install|uninstall|discover] [OPTIONS]
snyk-agent-scan guard

`guard install`

snyk-agent-scan guard install {claude,cursor,codex,all} [OPTIONS]

After configuring the hooks, installation sends a hooksConfiguredServerDiscovery event. It also configures a fire-and-forget session-start hook that reports discovered MCP servers with a sessionStartServerDiscovery event.

Flag	Type	Default	Description
`--url URL`	string	`https://api.snyk.io`	Remote hook base URL for the Snyk API environment.
`--tenant-id ID`	string	—	Snyk tenant UUID. Required when minting a push key; unnecessary when `PUSH_KEY` is set.
`--machine-id ID`	string	—	Required non-anonymous machine identifier sent in the `X-User` header's `identifier` field. May instead be set with `MACHINE_ID`.
`--file PATH`	string	—	Override the client configuration path.
`--managed`	boolean	`false`	Install in the admin/MDM-managed configuration rather than the user configuration.
`--test`	boolean	`false`	Deprecated/no-op.

`guard discover`

snyk-agent-scan guard discover [OPTIONS]

This internal command is invoked by the SessionStart hook configured by guard install. It reads the current target folder(s) from the selected client's hook payload, discovers MCP servers locally, and sends the resulting sessionStartServerDiscovery event directly to Agent Monitor; it is not normally run by hand.

Flag	Type	Default	Description
`--url URL`	string	`https://api.snyk.io`	Remote hook base URL for the Snyk API environment.
`--client {claude-code,cursor,codex}`	string	required	Hook client whose target-folder payload and endpoint conventions should be used.
`--scope {servers,skills,all}`	string	`all`	Discovery data to collect. The session-start hook installed by `guard install` passes `servers`, because the event it sends carries MCP servers only.

`guard uninstall`

snyk-agent-scan guard uninstall {claude,cursor,codex,all} [OPTIONS]

Flag	Type	Default	Description
`--file PATH`	string	—	Override the client configuration path.
`--managed`	boolean	`false`	Uninstall from the admin/MDM-managed configuration.

Guard environment variables

Variable	Purpose
`PUSH_KEY`	Pre-provisioned push key; skips minting when set
`TENANT_ID`	Tenant UUID alternative to `--tenant-id`
`SNYK_TOKEN`	Required to mint/revoke push keys and verify that Guard is enabled for the tenant
`MACHINE_ID`	Required non-anonymous machine identifier sent with hook events; alternative to `guard install --machine-id`
`AGENT_SCAN_COMMAND`	Optional Agent Scan command invoked by the session-start discovery hook, with the hook arguments appended. The hook is installed only when this is set. A value that is not an existing executable file is run as a shell command.

Environment variables

Variable	Used by	Description
`SNYK_TOKEN`	Standalone CLI	Snyk API token for analysis (`Authorization: token …`). Required for `scan` unless a push key is configured through `--push-key` (v0.6 and later) or the soon-to-be-deprecated `--control-server-H`. Never required for `evo`, which always mints and uses its own push key. Obtain it from https://app.snyk.io/account.
`SNYK_CLI_USE`	Binary invoked through Snyk CLI	Set to `true` by the extension so analysis uses the proxy-authenticated `/cli/analysis-machine` endpoint.
`SNYK_API`	Standalone CLI	Override the Snyk API base URL for local development or custom regions.
`SNYK_TENANT_ID`	Snyk CLI extension	Alternative to `--tenant-id` for the Go wrapper.
`AGENT_SCAN_ENVIRONMENT`	Analysis and upload	Environment label sent as `X-Environment` (default: `production`). Also accepts the older `MCP_SCAN_ENVIRONMENT` name.
`AGENT_SCAN_CI_HOSTNAME`	Upload	When `AGENT_SCAN_ENVIRONMENT=ci`, use this hostname instead of the machine node name.
`PUSH_KEY`	`guard install`	Pre-provisioned push key for hook installation.
`TENANT_ID`	`guard install`	Tenant UUID for hook installation.
`HOOK_VERSION`	Agent Guard hooks	Override the hook/API line version built into Guard artifacts.

Snyk CLI extension flags

The Snyk CLI command snyk agent-scan is implemented by cli-extension-agent-scan. It requires --experimental and delegates to the embedded Agent Scan binary.

Run snyk auth first. Analysis passes through a local credential proxy, so a valid Snyk CLI session is required even with --no-upload.

snyk agent-scan --experimental [EXTENSION_FLAGS] [AGENT_SCAN_ARGS...]

Extension-only flags

Flag	Type	Default	Description
`--experimental`	boolean	`false`	Required to enable the Agent Scan workflow.
`--tenant-id UUID`	string	—	Snyk tenant ID. Required with `--json` when no client ID is supplied; otherwise resolved interactively or from `SNYK_TENANT_ID`.
`--client-id UUID`	string	—	Push-key client ID passed to the control server as `x-client-id`. If omitted and upload is enabled, the extension mints one.
`--no-upload`	boolean	`false`	Do not upload results to Evo. Analysis still runs.
`--json`	boolean	`false`	Forward JSON output to the embedded binary.
`--skills [PATH]`	string/boolean	skills on	Using `--skills` alone is deprecated because skills are enabled by default. A supplied path remains supported and is forwarded as a scan target.

Snyk CLI flags consumed by the extension

Snyk CLI flag	Effect on Agent Scan
`--insecure`	Adds `--skip-ssl-verify` to the embedded binary
`--debug` / `-d`	Adds `--verbose` to the embedded binary

What the extension adds automatically

For a default scan without --no-upload, the extension:

Prepends scan if no path is supplied.
Sets --analysis-url from the configured Snyk API URL.
Mints or reuses a push key and appends the corresponding --control-server, --control-server-H, and --control-identifier arguments.
Starts a local proxy for Snyk CLI credential injection.

With --no-upload, it does not mint a client ID or add control-server arguments, but analysis still runs through the authenticated proxy. Explicit subcommands such as inspect, help, and guard do not configure upload.

The older snyk mcp-scan alias is deprecated and is treated as snyk agent-scan.

Examples

Standalone

# Full scan with v0.6 or later
uvx snyk-agent-scan@latest

# Full scan pinned to v0.5.x
uvx snyk-agent-scan@0.5.17

# Scan MCP components but skip skills
snyk-agent-scan --no-skills

# Scan a specific configuration or skill directory
snyk-agent-scan ~/.cursor/mcp.json
snyk-agent-scan ~/.claude/skills

# Inspect without analysis
snyk-agent-scan inspect

# v0.5.x CI with selected issue/failure codes ignored
snyk-agent-scan --ci --dangerously-run-mcp-servers \
  --ignore-issues-codes W001,X001

# v0.6+ CI with selected risks and failures ignored
snyk-agent-scan --ci --dangerously-run-mcp-servers \
  --ignore-risks dangerous_words,suspicious_download_url \
  --ignore-failure-codes X001

# Load flags from a YAML config file (CLI flags still override it)
snyk-agent-scan --config-file agent-scan.yaml
snyk-agent-scan --config-file agent-scan.yaml --server-timeout 5

# Enterprise upload with a push key
snyk-agent-scan scan \
  --control-server "https://api.snyk.io/hidden/mcp-scan/push?version=2025-08-28" \
  --control-server-H "x-client-id: <push-key>" \
  --control-identifier "$(hostname)"

Snyk CLI extension

# Authenticated scan with upload to Evo
snyk auth
snyk agent-scan --experimental

# Local analysis without Evo upload
snyk agent-scan --experimental --no-upload

# JSON output
snyk agent-scan --experimental --json --tenant-id "<tenant-uuid>"

# Scan a specific path
snyk agent-scan --experimental ~/.claude/skills

Agent Guard

# Check hook status
snyk-agent-scan guard

# Install for all supported clients
SNYK_TOKEN=... snyk-agent-scan guard install all --tenant-id "<tenant-uuid>" --machine-id "<machine-id>"

# Install through an MDM-managed configuration
PUSH_KEY=... snyk-agent-scan guard install cursor --managed --machine-id "<machine-id>"

# Uninstall
snyk-agent-scan guard uninstall all

Scanning overview — shared discovery and analysis workflow
Issue codes — v0.5.x security findings
Risk reference — v0.6+ security risk indicators and scores
Failure codes — operational discovery, inspection, and analysis failures
JSON output — versioned programmatic output schemas
Developer guide — entrypoints — standalone, Snyk CLI, and MDM paths