
tailsnitch v1.7
A security auditor for Tailscale configurations. Scans your tailnet for misconfigurations, overly permissive access controls, and security best practice violations.
Tailsnitch
A security auditor for Tailscale configurations. Tailsnitch scans your tailnet for 57 misconfigurations, overly permissive access controls, and security best practice violations.
Quick Start
# 1. Set your Tailscale API credentials
export TS_API_KEY="tskey-api-..."
# 2. Run audit
tailsnitch
# 3. See only high-severity findings
tailsnitch --severity high
# 4. Fix some issues ~interactively~ yolo mode
tailsnitch --fix
Installation
Download Pre-built Binary
Download the latest release from GitHub Releases.
macOS users: Remove quarantine attribute after download:
sudo xattr -rd com.apple.quarantine tailsnitch
Install via Go
go install github.com/Adversis/tailsnitch@latest
Build from Source
git clone https://github.com/Adversis/tailsnitch.git
cd tailsnitch
go build -o tailsnitch .
Authentication
Tailsnitch supports two authentication methods. OAuth is preferred when both are configured.
Option 1: OAuth Client (Recommended)
OAuth clients provide scoped, auditable access that doesn't expire when employees leave.
export TS_OAUTH_CLIENT_ID="..."
export TS_OAUTH_CLIENT_SECRET="tskey-client-..."
Create an OAuth client at: https://login.tailscale.com/admin/settings/oauth
Required scopes for read-only audit:
all:read covers everything. Granting scopes individually:
| Scope | Used for |
|---|---|
policy_file:read | Tailnet policy file — ACL-, NET-, SSH-* |
devices:core:read | Device list — DEV-, NET-, ACL-011 |
dns:read | DNS configuration — DNS-001, DEV-007 |
auth_keys:read | Machine auth keys — AUTH-*, ACL-011 |
feature_settings:read | Tailnet settings — DEV-008, DEV-009, DEV-014 |
logs:network:read | Network flow logging setting — LOG-001 |
networking_settings:read | HTTPS certificate setting — NET-004 |
log_streaming:read | Log stream destinations — LOG-002 |
webhooks:read | Webhook endpoints — LOG-005, LOG-012 |
oauth_keys:read | OAuth clients — LOG-006 |
users:read | User roles and status — USER-001, LOG-006 |
account_settings:read | Security contact — LOG-011 |
devices:posture_attributes:read | Posture integrations — DEV-014 |
Any scope you leave out only affects the checks that need it: those checks report that they could not read the setting rather than passing.
AUTH-005 and AUTH-006 read the tailnet's federated identities, which the admin
console calls trust credentials. They arrive from the same keys listing as auth
keys, so auth_keys:read is expected to cover them. That has not been confirmed
against a live tailnet. If the keys listing cannot be read, both checks report
not evaluated rather than passing. Whether a missing scope returns an error or
instead returns the listing with the identities filtered out is unconfirmed; if
it filters silently, AUTH-005 would report that no trust credentials exist and
AUTH-006 would find nothing to check.
Additional scopes for fix mode:
devices:core- Delete devices, modify tags (requires tag selection)auth_keys- Delete auth keys
Tailnet Lock
DEV-010 and DEV-012 report on Tailnet Lock, which the Tailscale API does not
expose as a tailnet setting. Devices locked out by it are visible through the
API, but determining whether lock is enabled needs the local tailscale CLI,
which reads the daemon on the machine running tailsnitch. When auditing another
tailnet with --tailnet, treat that part of the result accordingly. Use
--tailscale-path if the binary is in a non-standard location.
Option 2: API Key
API keys operate as the user who created them and inherit that user's permissions.
export TS_API_KEY="tskey-api-..."
Create an API key at: https://login.tailscale.com/admin/settings/keys
Usage Examples
Basic Audit
# Run full audit
tailsnitch
# Show passing checks too (verbose)
tailsnitch --verbose
# Output as JSON for processing
tailsnitch --json
# Audit a specific tailnet (when OAuth client has access to multiple)
tailsnitch --tailnet mycompany.com
Filter Results
# Only show critical and high severity issues
tailsnitch --severity high
# Filter by category
tailsnitch --category access # ACL issues
tailsnitch --category auth # Authentication & keys
tailsnitch --category device # Device security
tailsnitch --category network # Network exposure
tailsnitch --category ssh # SSH rules
tailsnitch --category log # Logging & admin
# Run specific checks only
tailsnitch --checks ACL-001,AUTH-001,DEV-010
tailsnitch --checks stale-devices,tailnet-lock-not-enabled
# List all available checks
tailsnitch --list-checks
Interactive Fix Mode
Fix mode allows you to remediate issues directly via the Tailscale API:
# Interactive fix mode
tailsnitch --fix
# Preview what would be fixed (dry run)
tailsnitch --fix --dry-run
# Auto-select safe fixes (still requires confirmation)
tailsnitch --fix --auto
# Disable audit logging of fix actions
tailsnitch --fix --no-audit-log
API-fixable items:
| Check | Action |
|---|---|
| AUTH-001, AUTH-002, AUTH-003 | Delete auth keys |
| DEV-002 | Remove tags from user devices |
| DEV-004 | Delete stale devices |
| DEV-005 | Authorize pending devices |
Fix mode also provides direct links to the admin console for issues that require manual intervention.
SOC 2 Evidence Export
Generate evidence reports for SOC 2 audits with Common Criteria (CC) control mappings:
# Export as JSON
tailsnitch --soc2 json > soc2-evidence.json
# Export as CSV (for spreadsheets)
tailsnitch --soc2 csv > soc2-evidence.csv
The SOC 2 report includes:
- Per-resource test results (each device, key, ACL rule tested individually)
- CC code mappings (CC6.1, CC6.2, CC6.3, CC6.6, CC7.1, CC7.2, etc.)
- Pass/Fail/N/A status for each control test
- Timestamp for audit trail
Example CSV output:
resource_type,resource_id,resource_name,check_id,check_title,cc_codes,status,details,tested_at
device,node123,prod-server,DEV-001,Tagged devices with key expiry disabled,CC6.1;CC6.3,PASS,Tags: [tag:server] key expiry enabled,2025-01-05T10:30:00Z
key,tskey-auth-xxx,tskey-auth-xxx,AUTH-001,Reusable auth keys exist,CC6.1;CC6.2;CC6.3,FAIL,Reusable key expires in 45 days,2025-01-05T10:30:00Z
Ignore Known Risks
Create a .tailsnitch-ignore file to suppress findings for known-accepted risks:
# .tailsnitch-ignore
# Ignore informational checks
ACL-008 # We intentionally don't use groups
ACL-009 # Legacy ACLs are fine for our use case
# Ignore specific medium checks with justification
DEV-006 # External devices are approved contractors
LOG-001 # Flow logs require Enterprise plan
# Ignore one item within a check, instead of muting the whole check
ACL-011:tag:monitoring # broad by design; every other tag is still checked
AUTH-001:tskey-auth-xxxx # rotates automatically via CI, tracked in TICKET-123
A line names either a whole check (ACL-011) or one item within it
(CHECK-ID:item, split on the first colon - the item itself may contain
colons). A per-item rule suppresses only that item: the check still runs and
still reports everything else it finds. Suppressing every flagged item never
turns a failing check into a passing one - the finding stays, downgraded to
Informational, so a suppressed finding never reads as a satisfied control.
Ignore file locations (checked in order):
.tailsnitch-ignorein current directory~/.tailsnitch-ignorein home directory
Because the first location is the working directory, an ignore file can come
from a repository rather than from you. Every run reports which file it used
and how many findings and items it suppressed, and --json records this in
the ignore_file and ignored fields (CHECK-ID for a whole check,
CHECK-ID:item for one suppressed item). Use --no-ignore to skip the file.
# Use a specific ignore file
tailsnitch --ignore-file /path/to/ignore
# Disable ignore file processing entirely
tailsnitch --no-ignore
JSON Export and Processing
# Export full report
tailsnitch --json > audit.json
# Extract failed checks as TSV
tailsnitch --json | jq -r '
.suggestions
| map(select(.pass == false))
| .[]
| [.id, .title, .severity, .remediation]
| @tsv
' > findings.tsv
# Summary by severity
tailsnitch --json | jq '
.suggestions
| map(select(.pass == false))
| group_by(.severity)
| map({severity: .[0].severity, count: length})
'
# List critical/high issues with admin links
tailsnitch --json | jq -r '
.suggestions
| map(select(.pass == false and (.severity == "CRITICAL" or .severity == "HIGH")))
| .[]
| "\(.id): \(.title)\n Fix: \(.fix.admin_url // "manual")\n"
'
Command Reference
| Flag | Description |
|---|---|
--json | Output as JSON |
--severity | Filter by minimum severity: critical, high, medium, low, info |
--category | Filter by category: access, auth, network, ssh, log, device, dns |
--checks | Run specific checks (comma-separated IDs or slugs) |
--list-checks | List all available checks and exit |
--tailnet | Specify tailnet to audit (default: from API key) |
--verbose | Show passing checks too |
--fix | Enable interactive fix mode |
--auto | Auto-select safe fixes (requires --fix) |
--dry-run | Preview fix actions without executing (requires --fix) |
--no-audit-log | Disable audit logging of fix actions |
--soc2 | Export SOC 2 evidence: json or csv |
--tailscale-path | Path to tailscale CLI (for Tailnet Lock checks) |
--timeout | Overall time budget for the audit (default 2m) |
--ignore-file | Path to ignore file |
--no-ignore | Disable ignore file processing |
--version | Show version information |
Security Checks
Tailsnitch performs 57 security checks across 7 categories. See docs/CHECKS.md for detailed documentation of each check.
Critical Severity
| ID | Check | Risk |
|---|---|---|
| ACL-001 | Default 'allow all' policy | All devices have unrestricted access |
| ACL-002 | SSH autogroup:nonroot misconfiguration | SSH as any non-root user |
| ACL-006 | tagOwners too broad | Privilege escalation via tags |
| ACL-007 | autogroup:danger-all usage | Access granted to external users |
High Severity
| ID | Check | Risk |
|---|---|---|
| ACL-011 | Tag reach crosses a trust boundary | A stolen reusable key mints a tag that reaches everything |
| AUTH-001 | Reusable auth keys | Unlimited device additions if stolen |
| AUTH-002 | Long expiry auth keys | Extended exposure window |
| AUTH-003 | Pre-authorized keys | Bypass device approval |
| AUTH-006 | Federated identity subject too broad | Any principal the issuer vouches for can mint the tag |
| DEV-001 | Tagged devices without key expiry | Indefinite access |
| DEV-002 | User devices tagged | Persist after user removal |
| DEV-010 | Tailnet Lock disabled | No protection against stolen keys |
| DEV-012 | Pending Tailnet Lock signatures | Unsigned nodes need review |
| NET-001 | Funnel exposure | Public internet access |
| NET-003 | Subnet router trust boundary | Unencrypted traffic on local network |
| SSH-002 | Root SSH without check mode | No re-authentication required |
Medium Severity
| ID | Check | Risk |
|---|---|---|
| ACL-004 | autogroup:member usage | External users included |
| ACL-005 | AutoApprovers configured | Bypass route approval |
| AUTH-004 | Non-ephemeral CI/CD keys | Stale devices accumulate |
| AUTH-005 | Workload identity federation not in use | Long-lived keys stay stealable |
| DEV-003 | Outdated clients | Potential vulnerabilities |
| DEV-004 | Stale devices | Unused attack surface |
| DEV-005 | Unauthorized devices | Pending approval queue |
| DEV-007 | Sensitive machine names | CT log exposure |
| DEV-009 | Device approval config | May not be enabled |
| NET-004 | HTTPS CT log exposure | Machine names public |
| NET-005 | Exit node traffic visibility | Operator sees all traffic |
| NET-006 | Serve exposure | Local services on tailnet |
| SSH-003 | Recorder UI exposure | Sessions visible to network |
Informational
Checks for logging configuration, DNS settings, user roles, and manual verification items.
Severity that depends on the finding
Several checks rate what they find rather than carrying one fixed severity. Three are worth calling out here:
- ACL-011 reports every tag's reach at Informational. It fails only when a tag an auth key can assign reaches a wildcard destination, a routed subnet or exit-node egress: High if a reusable key assigns that tag, Medium if only a one-off key does. A device count never sets severity.
- AUTH-005 reports Medium when the tailnet has no trust credentials at all, and Low when trust credentials exist but a reusable key still mints tags none of them cover.
- AUTH-006 reports High for a subject that is nothing but a wildcard, and Low for a narrower wildcard or a missing audience.
Output Example
+=====================================================================+
| TAILSNITCH SECURITY AUDIT |
| Tailnet: example.com |
| Version: 1.0.0 (build: abc123) |
+=====================================================================+
Using ignore file: .tailsnitch-ignore (3 rules)
=== ACCESS CONTROLS ===================================================
[CRITICAL] ACL-001: Default 'allow all' policy active
Your ACL policy omits the 'acls' field. Tailscale applies a
default 'allow all' policy, granting all devices full access.
Remediation:
Define explicit ACL rules following least privilege principle.
Source: https://tailscale.com/docs/reference/examples/acls
----------------------------------------------------------------------
=== AUTHENTICATION & KEYS =============================================
[HIGH] AUTH-001: Reusable auth keys exist
Found 2 reusable auth key(s). These can be reused to add
multiple devices if compromised.
Details:
- Key tskey-auth-xxx (expires in 45 days)
- Key tskey-auth-yyy (expires in 89 days)
Remediation:
Store reusable keys in a secrets manager. Prefer one-off keys.
Source: https://tailscale.com/docs/features/access-control/auth-keys
----------------------------------------------------------------------
SUMMARY
======================================================================
Critical: 1 High: 3 Medium: 5 Low: 2 Info: 8
Total findings: 19 | Passed: 33
Tailnet Lock Checks
Tailnet Lock checks (DEV-010, DEV-012) require the local tailscale CLI and run against the local machine's daemon. When auditing a remote tailnet via --tailnet, these checks reflect local status, not the audited tailnet.
# Specify custom tailscale binary path if needed
tailsnitch --tailscale-path /opt/tailscale/bin/tailscale
CI/CD Integration
Run Tailsnitch in CI/CD pipelines to catch security regressions:
# GitHub Actions example
- name: Audit Tailscale Security
env:
TS_OAUTH_CLIENT_ID: ${{ secrets.TS_OAUTH_CLIENT_ID }}
TS_OAUTH_CLIENT_SECRET: ${{ secrets.TS_OAUTH_CLIENT_SECRET }}
run: |
tailsnitch --json > audit.json
# Fail if critical or high severity issues exist
if tailsnitch --severity high --json | jq -e '.summary.critical + .summary.high > 0' > /dev/null; then
echo "Critical or high severity issues found!"
tailsnitch --severity high
exit 1
fi
References
License
MIT
Contributing
See CONTRIBUTING.md for guidelines.