@cclabsnz/sf-audit 1.8.1 → 1.9.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (77) hide show
  1. package/CHANGELOG.md +199 -0
  2. package/README.md +138 -758
  3. package/lib/chains/CapabilityRegistry.d.ts +6 -2
  4. package/lib/chains/CapabilityRegistry.js +48 -2
  5. package/lib/chains/namedChains.js +140 -7
  6. package/lib/checks/impl/AgentActionSurfaceCheck.d.ts +1 -1
  7. package/lib/checks/impl/AgentChannelExposureCheck.d.ts +1 -1
  8. package/lib/checks/impl/AgentChannelExposureCheck.js +52 -10
  9. package/lib/checks/impl/AgentInventoryCheck.d.ts +1 -1
  10. package/lib/checks/impl/AgentMonitoringCoverageCheck.d.ts +1 -1
  11. package/lib/checks/impl/AgentUserPrivilegeCheck.d.ts +1 -1
  12. package/lib/checks/impl/ApexCrudFLSCheck.d.ts +1 -1
  13. package/lib/checks/impl/ApexCrudFLSCheck.js +3 -3
  14. package/lib/checks/impl/ApexLoggingCheck.d.ts +1 -1
  15. package/lib/checks/impl/ApexRestEndpointCheck.d.ts +1 -1
  16. package/lib/checks/impl/ApexRestEndpointCheck.js +0 -3
  17. package/lib/checks/impl/ApexSharingCheck.d.ts +1 -1
  18. package/lib/checks/impl/CodeSecurityCheck.d.ts +1 -1
  19. package/lib/checks/impl/ConnectedAppInactivityCheck.d.ts +1 -1
  20. package/lib/checks/impl/ConnectedAppsCheck.d.ts +1 -1
  21. package/lib/checks/impl/CspTrustedSitesCheck.d.ts +1 -1
  22. package/lib/checks/impl/CustomLabelsCredentialCheck.js +1 -1
  23. package/lib/checks/impl/DataClassificationCheck.js +15 -1
  24. package/lib/checks/impl/DeploymentIdentityCheck.d.ts +1 -1
  25. package/lib/checks/impl/EventMonitoringCheck.d.ts +1 -1
  26. package/lib/checks/impl/FailedLoginCheck.js +16 -1
  27. package/lib/checks/impl/FieldLevelSecurityCheck.js +29 -16
  28. package/lib/checks/impl/GuestExecutableApexCheck.d.ts +1 -1
  29. package/lib/checks/impl/GuestTrafficAnomalyCheck.d.ts +1 -1
  30. package/lib/checks/impl/GuestUserAccessCheck.d.ts +1 -1
  31. package/lib/checks/impl/GuestUserVisibilityCheck.d.ts +35 -0
  32. package/lib/checks/impl/GuestUserVisibilityCheck.js +168 -0
  33. package/lib/checks/impl/HardcodedCredentialsCheck.d.ts +2 -2
  34. package/lib/checks/impl/HardcodedCredentialsCheck.js +0 -1
  35. package/lib/checks/impl/HealthCheckCheck.d.ts +1 -1
  36. package/lib/checks/impl/HealthCheckCheck.js +1 -3
  37. package/lib/checks/impl/IntegrationLeastPrivilegeCheck.d.ts +42 -0
  38. package/lib/checks/impl/IntegrationLeastPrivilegeCheck.js +467 -0
  39. package/lib/checks/impl/IntegrationUsersCheck.js +48 -52
  40. package/lib/checks/impl/InternalUserMfaCheck.js +0 -1
  41. package/lib/checks/impl/IpRestrictionsCheck.js +39 -5
  42. package/lib/checks/impl/MfaMethodStrengthCheck.d.ts +1 -1
  43. package/lib/checks/impl/MfaRegistrationCheck.d.ts +1 -1
  44. package/lib/checks/impl/NamedCredentialsCheck.d.ts +1 -1
  45. package/lib/checks/impl/PrivilegedAccessCheck.d.ts +1 -1
  46. package/lib/checks/impl/PublicGroupSharingCheck.d.ts +1 -1
  47. package/lib/checks/impl/PublicGroupSharingCheck.js +21 -2
  48. package/lib/checks/impl/RemoteSitesCheck.d.ts +1 -1
  49. package/lib/checks/impl/ScheduledApexCheck.d.ts +1 -1
  50. package/lib/checks/impl/ScheduledApexCheck.js +0 -1
  51. package/lib/checks/impl/SeparationOfDutiesCheck.d.ts +1 -1
  52. package/lib/checks/impl/SessionHardeningCheck.d.ts +1 -1
  53. package/lib/checks/impl/SiemIntegrationCheck.d.ts +1 -1
  54. package/lib/checks/impl/SoapLoginApiAuthCheck.d.ts +10 -0
  55. package/lib/checks/impl/SoapLoginApiAuthCheck.js +151 -0
  56. package/lib/checks/impl/SsoEnforcementCheck.js +0 -1
  57. package/lib/checks/impl/TrustedUrlHygieneCheck.d.ts +1 -1
  58. package/lib/checks/impl/TrustedUrlHygieneCheck.js +3 -1
  59. package/lib/checks/impl/VisualforceXssCheck.d.ts +1 -1
  60. package/lib/checks/registry.js +4 -0
  61. package/lib/checks/support/integrationAccounts.d.ts +35 -0
  62. package/lib/checks/support/integrationAccounts.js +209 -0
  63. package/lib/commands/audit/security.d.ts +12 -1
  64. package/lib/commands/audit/security.js +32 -9
  65. package/lib/compliance/catalogs/gdpr.d.ts +3 -0
  66. package/lib/compliance/catalogs/gdpr.js +54 -0
  67. package/lib/compliance/catalogs/hipaa.d.ts +3 -0
  68. package/lib/compliance/catalogs/hipaa.js +97 -0
  69. package/lib/compliance/catalogs/index.js +4 -0
  70. package/lib/compliance/mapping.js +111 -36
  71. package/lib/findings/CheckMeta.js +2 -0
  72. package/lib/findings/exitCode.d.ts +28 -0
  73. package/lib/findings/exitCode.js +46 -0
  74. package/lib/findings/loadScoringConfig.js +5 -5
  75. package/lib/renderers/HtmlRenderer.js +1 -1
  76. package/messages/audit.security.md +6 -0
  77. package/package.json +16 -11
package/README.md CHANGED
@@ -8,15 +8,16 @@
8
8
  [![CodeQL](https://github.com/cclabsnz/sf-audit-plugin/actions/workflows/codeql.yml/badge.svg)](https://github.com/cclabsnz/sf-audit-plugin/actions/workflows/codeql.yml)
9
9
  [![Semgrep](https://github.com/cclabsnz/sf-audit-plugin/actions/workflows/semgrep.yml/badge.svg)](https://github.com/cclabsnz/sf-audit-plugin/actions/workflows/semgrep.yml)
10
10
  [![OpenSSF Scorecard](https://api.securityscorecards.dev/projects/github.com/cclabsnz/sf-audit-plugin/badge)](https://securityscorecards.dev/viewer/?uri=github.com/cclabsnz/sf-audit-plugin)
11
+ [![OpenSSF Best Practices](https://www.bestpractices.dev/projects/14149/badge)](https://www.bestpractices.dev/projects/14149)
11
12
  [![npm version](https://img.shields.io/npm/v/@cclabsnz/sf-audit)](https://www.npmjs.com/package/@cclabsnz/sf-audit)
12
13
  [![npm provenance](https://img.shields.io/badge/npm-signed%20provenance-brightgreen)](https://www.npmjs.com/package/@cclabsnz/sf-audit#provenance)
13
14
  [![License: Apache-2.0](https://img.shields.io/badge/license-Apache--2.0-blue.svg)](LICENSE)
14
15
 
15
16
  A Salesforce CLI (`sf`) plugin that runs a complete, **read-only** security audit against any Salesforce org, risk-scores it with an A–F grade, and turns the result into a report your security team (or your client's) can act on.
16
17
 
17
- - **88 read-only checks** across identity, access, data, code, integrations, monitoring, and Agentforce / GenAI
18
- - **Attack-chain correlation:** links individual findings into named, multi-step attack scenarios
19
- - **Compliance mapping:** every finding mapped to **source-verified** controls across 8 frameworks (OWASP, OWASP LLM Top 10, SOC 2, ISO/IEC 27001:2022, Security Benchmark for Salesforce, NZ Privacy Act, HISO 10029, NZISM)
18
+ - **90 read-only checks** across identity, access, data, code, integrations, monitoring, and Agentforce / GenAI
19
+ - **Attack-chain correlation:** links individual findings into named, multi-step attack scenarios — eleven modelled chains, plus an emergent pass for combinations nobody has named yet (see [Attack chains](#attack-chains))
20
+ - **Compliance mapping:** every finding mapped to **source-verified** controls across 10 frameworks (OWASP, OWASP LLM Top 10, SOC 2, ISO/IEC 27001:2022, Security Benchmark for Salesforce, NZ Privacy Act, HISO 10029, NZISM, HIPAA Security Rule, GDPR)
20
21
  - **Outputs:** a technical `html` / `md` / `json` report, or a branded, client-ready **executive report** (print-to-PDF) with priorities, remediation roadmap, and a compliance coverage matrix
21
22
  - **History & diff:** archives each run and shows security-posture drift over time
22
23
  - **Free event baseline:** `sf audit events pull` captures the org's free daily `EventLogFile` logs to local disk before the 1-day retention window drops them — no Event Monitoring / Shield add-on needed
@@ -29,33 +30,13 @@ A Salesforce CLI (`sf`) plugin that runs a complete, **read-only** security audi
29
30
  sf plugins install @cclabsnz/sf-audit
30
31
  ```
31
32
 
32
- **You will be warned that this plugin is not digitally signed.** That is expected, and it is not a judgement on this plugin. Salesforce's signature verification only accepts public keys served from `developer.salesforce.com`, so no third-party plugin can satisfy it: every community plugin in the ecosystem produces the same prompt. Answer `y` to continue.
33
+ You will be warned that the plugin is not digitally signed. That is expected and says nothing about
34
+ this plugin: Salesforce only accepts signing keys served from `developer.salesforce.com`, so no
35
+ third-party plugin can satisfy it. Answer `y`. For unattended installs, allowlisting and local
36
+ development setup are in **[docs/INSTALL.md](docs/INSTALL.md)**.
33
37
 
34
- Since "trust us" is a poor answer from a tool that authenticates against your production org, the guarantees this project offers are ones you can check yourself rather than take on faith. Releases carry signed npm provenance, so you can confirm the published tarball was built from the exact public commit, and the read-only promise is enforced by a test that fails the build if a single write path appears in the source:
35
-
36
- ```bash
37
- npm audit signatures # reports "verified attestations" for @cclabsnz/sf-audit
38
- ```
39
-
40
- See [Trust & verification](#trust--verification) for the full list, including the no-network-egress guard and the independent scans.
41
-
42
- For CI or any unattended install, where an interactive prompt would hang the job, allowlist the package on the machine doing the installing:
43
-
44
- ```jsonc
45
- // macOS and Linux: ~/.config/sf/unsignedPluginAllowList.json
46
- // Windows: %LOCALAPPDATA%\sf\unsignedPluginAllowList.json
47
- ["@cclabsnz/sf-audit"]
48
- ```
49
-
50
- Or, for local development:
51
-
52
- ```bash
53
- git clone https://github.com/cclabsnz/sf-audit-plugin.git
54
- cd sf-audit-plugin
55
- npm install
56
- npm run build
57
- sf plugins link .
58
- ```
38
+ Since "trust us" is a poor answer from a tool that authenticates against your production org, the
39
+ guarantees here are checkable rather than asserted — see [Trust & verification](#trust--verification).
59
40
 
60
41
  ## Usage
61
42
 
@@ -63,253 +44,108 @@ sf plugins link .
63
44
  sf audit security --target-org <orgAlias>
64
45
  ```
65
46
 
66
- This runs all 88 security checks against the target org and writes a report to the current directory.
47
+ Runs all 90 checks and writes an HTML report to the current directory. `sf audit list` prints every
48
+ check id.
67
49
 
68
- List every available check id (the values you pass to `--checks`):
50
+ | Flag | Default | |
51
+ |------|---------|---|
52
+ | `--format` / `-f` | `html` | `html`, `md`, `json`, `executive` — comma-separated |
53
+ | `--fail-on` | (none) | Exit 1 if any finding is at or above `CRITICAL`/`HIGH`/`MEDIUM`/`LOW`. For CI |
54
+ | `--fail-on-inconclusive` | `false` | Exit 3 if any check could not gather evidence. For CI |
55
+ | `--checks` | *(all)* | Run only these check ids |
56
+ | `--frameworks` | `universal` | Compliance matrix scope for the executive report |
69
57
 
70
58
  ```bash
71
- sf audit list
72
- ```
73
-
74
- ### Options
75
-
76
- | Flag | Default | Description |
77
- |------|---------|-------------|
78
- | `--target-org` | *(required)* | Org alias or username to audit |
79
- | `--format` / `-f` | `html` | Output format(s), comma-separated: `html`, `md`, `json`, `executive` |
80
- | `--output` / `-o` | `.` | Directory to write the report file |
81
- | `--fail-on` | (none) | Exit with code 1 if any finding is at or above this severity: `CRITICAL`, `HIGH`, `MEDIUM`, `LOW` |
82
- | `--checks` | *(all)* | Comma-separated check IDs to run instead of all 88 (e.g. `hardcoded-credentials,apex-sharing`) |
83
- | `--scoring-config` | (none) | Path to a custom scoring config JSON file to override weights and grade thresholds |
84
- | `--prepared-for` | (none) | Client name for the executive report cover line |
85
- | `--branding` | (none) | Path to a `report-branding.json` to override CloudCounsel defaults (executive format) |
86
- | `--top` | `5` | Number of executive priorities to highlight (executive format) |
87
- | `--frameworks` | `universal` | Compliance matrix scope (executive format): `universal` (OWASP/OWASP LLM/SOC 2/ISO 27001), `nz` (ISO/HISO/Privacy Act/NZISM), `all`, or a comma list (e.g. `owasp,owasp-llm,iso,nzism`) |
88
- | `--resolve-domains` | `false` | Makes **outbound DNS queries from this machine** to verify CSP trusted domains still resolve (flags unresolvable / parked domains as exfiltration channels). Off by default; a default run contacts **only the target org** and never reaches out to any other host. |
89
-
90
- ### Examples
91
-
92
- ```bash
93
- # HTML report (default)
94
- sf audit security --target-org myOrg
95
-
96
- # Multiple formats at once
97
- sf audit security --target-org myOrg --format html,md,json
98
-
99
- # Write report to a specific directory
100
- sf audit security --target-org myOrg --output ./reports
101
-
102
- # Fail CI pipeline on HIGH or CRITICAL findings
59
+ # Fail a pipeline on HIGH or worse
103
60
  sf audit security --target-org myOrg --fail-on HIGH
104
61
 
105
- # Run only specific checks
106
- sf audit security --target-org myOrg --checks hardcoded-credentials,apex-sharing,guest-user-access
62
+ # Also fail if the audit user could not see enough to judge
63
+ sf audit security --target-org myOrg --fail-on HIGH --fail-on-inconclusive
107
64
 
108
- # Guest / Experience Cloud exposure sweep — the unauthenticated data-leak surface
109
- # (bulk-read via UI API, guest-owned records defeating Private OWD, file access, self-reg, threat detection)
110
- sf audit security --target-org myOrg --checks guest-user-access,guest-object-exposure,guest-site-options,guest-executable-apex,experience-cloud-site,threat-detection
65
+ # Branded, client-ready PDF-able report
66
+ sf audit security --target-org myOrg --format executive --prepared-for "Acme Health"
111
67
 
112
- # Use a custom scoring config (e.g. stricter weights for your org)
113
- sf audit security --target-org myOrg --scoring-config ./my-scoring.json
68
+ # Machine-readable: result on stdout, progress logging suppressed
69
+ sf audit security --target-org myOrg --json
114
70
  ```
115
71
 
116
- ### Executive report
72
+ ### Exit codes
117
73
 
118
- `--format executive` produces a CloudCounsel-branded, print-to-PDF HTML report for clients:
119
- grade and executive summary, top priorities with abuse/impact narratives, attack scenarios, a
120
- risk×effort remediation roadmap, and a **compliance coverage matrix** mapping findings to framework
121
- controls. It is fully self-contained (fonts embedded); open it and **Save as PDF**.
74
+ | Code | Meaning |
75
+ |-----:|---------|
76
+ | `0` | Audit completed; nothing the caller asked to fail on |
77
+ | `1` | Findings at or above `--fail-on` |
78
+ | `2` | The audit could not run — authentication, connection, or bad flags |
79
+ | `3` | Audit ran, but checks could not gather evidence (`--fail-on-inconclusive` only) |
122
80
 
123
- ```bash
124
- # Branded executive report for a client (universal compliance matrix)
125
- sf audit security --target-org myOrg --format executive --prepared-for "Acme Health" --top 5
81
+ A definite finding outranks unknown coverage: when both apply you get `1`, and the
82
+ inconclusive count is still in the report body. `3` exists so a pipeline can tell
83
+ "this org is fine" apart from "the audit user could not see enough to judge".
126
84
 
127
- # NZ health/government engagement: NZ framework matrix
128
- sf audit security --target-org myOrg --format executive --frameworks nz
85
+ All twelve flags, more examples, and what the executive report contains:
86
+ **[docs/COMMANDS.md](docs/COMMANDS.md)**.
129
87
 
130
- # White-label / co-brand via overrides
131
- sf audit security --target-org myOrg --format executive --branding ./report-branding.json
132
- ```
88
+ ## What It Checks
133
89
 
134
- Compliance controls are mapped from authoritative, version-pinned sources (OWASP Top 10:2021,
135
- OWASP Top 10 for LLM Applications 2025, AICPA TSC, ISO/IEC 27001:2022, the Security Benchmark for
136
- Salesforce, NZ Privacy Act, HISO 10029, NZISM). Only source-verified controls render. "No findings detected" is **not** an attestation of
137
- compliance (see the report's Scope & Liability section).
90
+ **90 read-only checks** across ten domains. Every finding is risk-rated CRITICAL → INFO, mapped to
91
+ compliance controls, and correlated into [attack chains](#attack-chains).
138
92
 
139
- The report file is written as `sf-audit-<orgId>-<timestamp>.<ext>` in the output directory (e.g. `sf-audit-00D000000000001-1711234567890.html`).
93
+ | Domain | Checks |
94
+ |--------|-------:|
95
+ | Identity & Authentication | 13 |
96
+ | Users, Permissions & Privilege | 12 |
97
+ | Data Access & Sharing | 11 |
98
+ | Guest & External-Facing Access | 12 |
99
+ | Integrations, Connected Apps & Deployments | 10 |
100
+ | Monitoring & Threat Detection | 10 |
101
+ | Apex & Code Security | 9 |
102
+ | AI & Agents (Agentforce / GenAI) | 6 |
103
+ | Org Health & Configuration | 5 |
104
+ | Secrets & Credential Storage | 2 |
140
105
 
141
- ## What It Checks
106
+ Full inventory, with what each check looks for: **[docs/CHECKS.md](docs/CHECKS.md)**.
142
107
 
143
- The audit runs **88 read-only checks**. Every finding is risk-rated (CRITICAL → INFO) and mapped to controls across the compliance frameworks (see [Compliance frameworks](#compliance-frameworks)). The checks are grouped into ten domains below.
144
-
145
- ### Org Health & Configuration
146
- | Check | What it looks for |
147
- |-------|------------------|
148
- | Security Health Check | Salesforce Health Check score and individual high-risk settings |
149
- | Enhanced Domains | Enhanced Domains enabled: prevents cross-org cookie leakage and enforces URL isolation |
150
- | Pending Release Updates | Salesforce release updates pending activation, especially those past auto-activation |
151
- | Legacy API Versions | Apex compiled on old API versions and SOAP-based remote site integrations |
152
- | API & Resource Limits | API request consumption against daily and concurrent limits |
153
-
154
- ### Identity & Authentication
155
- | Check | What it looks for |
156
- |-------|------------------|
157
- | SSO Enforcement | Username-password logins indicating SSO is not org-wide enforced |
158
- | My Domain Login Policy | My Domain configured and login from login.salesforce.com blocked (stops SSO bypass) |
159
- | Internal User MFA | MFA enforcement for active internal standard users |
160
- | MFA for External Users | MFA enforced for external/portal users with data access |
161
- | MFA Method Registration | Active standard users with no registered MFA method |
162
- | MFA Method Strength | Registered MFA methods classified by strength (phishing-resistant / TOTP / weak) |
163
- | High Assurance Sessions | Admin-capable connected apps requiring short timeouts or high-assurance MFA sessions |
164
- | Trusted IP Ranges | Trusted IP ranges that bypass MFA, including overly broad ranges |
165
- | Login IP Restrictions | Admin profiles missing IP ranges; connected apps with relaxed IP policy |
166
- | Password & Session Policy | Password complexity, session timeout, and MFA gaps (from Health Check) |
167
- | Certificate Expiry | Installed certificates nearing expiry (30 / 90 / 180-day thresholds) |
168
- | Auth Providers & External IdPs | External Auth Providers and SAML SSO configs; flags social/JIT providers that can federate in or auto-provision users |
169
- | Session & Clickjack Hardening | Clickjack, CSRF, XSS, and content-sniffing settings read authoritatively from SecuritySettings (Health Check cache fallback) with per-setting remediation |
170
-
171
- ### Users, Permissions & Privilege
172
- | Check | What it looks for |
173
- |-------|------------------|
174
- | Users & Admins | Users with system-wide permissions (ModifyAllData, ViewAllData, AuthorApex, CustomizeApplication) |
175
- | Permissions | Unassigned permission sets and high profile counts that widen the attack surface |
176
- | Standard Profile Usage | Active users assigned to out-of-the-box standard profiles |
177
- | Use Any API Client | Users with the permission that bypasses API Access Control |
178
- | Privilege Escalation Permissions | Users holding lateral-movement / persistence permission clusters |
179
- | Privileged Access & Shadow Admins | Effective high-risk permissions per user (profile + permission sets + groups); admin-equivalent users not on the System Administrator profile |
180
- | Separation of Duties | Toxic permission combinations a single user holds (e.g. Manage Users + Assign Permission Sets, Author Apex + Modify All Data) |
181
- | Integration / Service Accounts | Non-human identity inventory and excess privilege |
182
- | Inactive Users | Active licensed users with no login in 90+ days |
183
- | Login-As & Delegated Administration | Delegated-admin groups (SOQL) and the "log in as any user" policy read from SecuritySettings — user-impersonation and scoped-escalation paths |
184
- | Mass Data Export Access | Profiles/permission sets with Weekly Data Export, or API access combined with View/Modify All Data (bulk-exfil capability) |
185
-
186
- ### Data Access & Sharing
187
- | Check | What it looks for |
188
- |-------|------------------|
189
- | OWD Sharing Model | Org-wide defaults for Account, Contact, Opportunity, Case, Lead (internal + external) |
190
- | Field-Level Security | Sensitive fields (SSN, credit card, tax ID) exposed to broad permission sets |
191
- | Public Group Sharing | Sharing rules that grant access to All Internal Users |
192
- | Report Folder Public Access | Report folders any authenticated user can view |
193
- | Field History Tracking | History tracking enabled on sensitive standard objects |
194
- | Data Classification & Encryption | Field data classification usage and Shield Platform Encryption |
195
- | Encryption Coverage for Sensitive Fields | Fields classified PII/PHI (ComplianceGroup) that are NOT encrypted at rest with Shield |
196
- | Sandbox Data Masking | In sandboxes, populated PII fields (likely unmasked production data); advises running Data Mask |
197
- | Flows Without Sharing | Active flows running in system context without sharing enforcement |
198
- | Content Distribution Links | Public file links missing expiry or passwords, and stale records |
199
- | Public Static Resources & Documents | Documents marked externally available (anonymous URL access) and static resources cached publicly |
200
-
201
- ### Guest & External-Facing Access
202
- | Check | What it looks for |
203
- |-------|------------------|
204
- | Guest User Access | Object permissions and sharing rules granted to unauthenticated guests |
205
- | Guest Object Exposure (Bulk Read via UI API) | Auto-discovers guest-readable objects (incl. records exposed by guest ownership despite a Private OWD), then grades them by actual UI-API reachability: objects the UI API models are CRITICAL (GraphQL-pullable), objects readable only in the sharing model but not UI-API-modelled (Calendar, AuthSession, etc.) are separated as MEDIUM, with UserRecordAccess as ground-truth read confirmation |
206
- | Guest-Executable Apex | Apex that guest profiles can run, flagging `without sharing` classes |
207
- | Experience Cloud Sites | Live sites with self-registration enabled and guest user presence |
208
- | Experience Cloud Guest Site Options | Guest file access and guest member visibility on Experience Cloud sites |
209
- | Secure Guest User Record Access | Verifies the "Secure guest user record access" enforcement is activated — the guardrail that stops guest-owned records defeating a Private OWD (complements Guest Object Exposure) |
210
- | Guest API & Bulk Access | Guest users granted API Enabled or Bulk API Hard Delete — programmatic bulk read/delete for unauthenticated visitors |
211
- | Classic Force.com Sites | Active classic (Visualforce) Sites and their guest users — an unauthenticated surface separate from Experience Cloud |
212
- | Experience Cloud CSP & Lightning Web Security | Advises verifying Strict CSP and Lightning Web Security on live sites (not reliably API-readable) |
213
- | CORS Allowlist | Wildcard or overly broad CORS allowlist origins |
214
- | CSP Trusted Sites | Content Security Policy trusted sites with insecure HTTP (mixed-content) endpoints |
215
-
216
- ### Apex & Code Security
217
- | Check | What it looks for |
218
- |-------|------------------|
219
- | Apex Sharing Declarations | Classes classified by sharing declaration (with / without / inherited / omitted) |
220
- | Apex CRUD/FLS Enforcement | DML or SOQL performed without CRUD/FLS permission checks |
221
- | Apex REST Endpoints | `@RestResource` classes running `without sharing` |
222
- | Visualforce XSS | `escape="false"` and unencoded merge fields in Visualforce markup |
223
- | Hardcoded Credentials | Bearer tokens, Basic auth, API keys, and raw callout URLs in Apex |
224
- | Code Security & Coverage | Org-wide Apex test coverage, class/trigger counts, and SOQL injection patterns |
225
- | Scheduled & Batch Apex | Active scheduled and batch Apex jobs |
226
- | Anonymous Apex Audit | Anonymous Apex executed in the last 90 days (via SetupAuditTrail) |
227
- | Apex Logging Framework | Persistent logging usage and sensitive data exposed in Apex logs |
228
-
229
- ### Integrations, Connected Apps & Deployments
230
- | Check | What it looks for |
231
- |-------|------------------|
232
- | Connected Apps | Apps not restricted to admin-approved users |
233
- | Connected App OAuth Scopes | Full OAuth-scope grants and infinite refresh-token policies |
234
- | Inactive Connected Apps | Apps with no OAuth logins in the past 90 days |
235
- | Named Credentials | Named credential inventory; credentials not referenced in Apex |
236
- | External Credential Authentication | External Credentials using no authentication or a custom (non-standard) scheme |
237
- | Remote Site Settings | Remote sites with protocol security disabled |
238
- | Outbound Messages | Workflow outbound messages that include a session ID or post to cleartext (http://) endpoints |
239
- | Email Security & Spoofing | Inbound email services accepting mail from any sender / running Apex unauthenticated, and org-wide send-as addresses open to all profiles |
240
- | Installed Packages | Managed/unmanaged package inventory; unmanaged or beta packages in production |
241
- | Deployment Identity | Designated deployment identity and uncontrolled deployment activity |
242
-
243
- ### Secrets & Credential Storage
244
- | Check | What it looks for |
245
- |-------|------------------|
246
- | Custom Settings & Credentials | Custom settings with credential-like names that may store secrets |
247
- | Custom Labels Credential Exposure | API keys and tokens in globally-readable Custom Labels |
248
-
249
- ### Monitoring & Threat Detection
250
- | Check | What it looks for |
251
- |-------|------------------|
252
- | Audit Trail | Permission changes and Login-As events in the setup audit trail |
253
- | Login Session | Failed login trends, Login-As events, and access from diverse IPs |
254
- | Failed Login Detection | Brute-force and credential-stuffing patterns (last 7 days) |
255
- | Transaction Security Policies | Automated threat detection and response policies configured |
256
- | Active Debug Log Traces | Active TraceFlag records capturing logs, including high-detail traces |
257
- | Event Monitoring | Event Monitoring enabled with logs covering 30+ days |
258
- | Threat Detection Event Storage | Guest User Anomaly / Threat Detection event storage enabled and retaining events |
259
- | Guest Traffic Anomaly | Scans recent EventLogFile guest requests for anonymizer/hosting source IPs, single-IP bursts, and GraphQL `totalCount` reconnaissance sweeps |
260
- | Anomalous Successful Logins | Accounts with successful logins from an unusually high number of distinct source IPs (credential-sharing / compromise signature) |
261
- | SIEM Integration Signals | Evidence of SIEM or external monitoring integration |
262
-
263
- ### AI & Agents (Agentforce / GenAI)
264
- | Check | What it looks for |
265
- |-------|------------------|
266
- | Agent Inventory | Every Agentforce agent, its active version, and its run-as user (inventory findings); flags an active agent whose run-as identity is inactive or frozen |
267
- | Agent User Privilege | Agent / run-as users with Modify All Data or View All Data, broad object write access, or read access to objects classified as sensitive — the data a prompt injection inherits |
268
- | Agent Action Surface | Write-capable agent actions (Apex/Flow that create, update, or delete) and agents with an unusually large action surface |
269
- | Agentforce Channel Exposure | Correlates active agents with the channels that reach them (Experience Cloud sites, embedded deployments, messaging channels); flags guest-reachable exposure |
270
- | Agentforce Monitoring Coverage | Active agents running with no Event Monitoring capture and no Transaction Security policy (the monitoring gap in the ForcedLeak pattern); points at `sf audit events pull` |
271
- | Trusted URL Hygiene | Reviews the CSP trusted-sites allowlist for non-Salesforce domains that could be repurposed as exfiltration channels; with `--resolve-domains`, DNS-checks each for unresolvable or parked entries |
272
-
273
- Two named attack chains correlate these findings: **Prompt injection blast radius** (guest-reachable channel + over-privileged agent user + write-capable actions) and **ForcedLeak pattern** (active agents + a stale/unresolvable trusted URL + no event capture).
274
-
275
- The five agent-specific checks stay silent in orgs where Agentforce is not enabled (the GenAI objects do not exist, so the inventory records `not-enabled` and the dependent checks return nothing). Trusted URL Hygiene runs everywhere, since the CSP allowlist is an org-wide exfiltration surface regardless of Agentforce.
276
-
277
- ### Known limitations (advisory-only checks)
278
-
279
- A small number of checks emit an **advisory** rather than a pass/fail verdict because the underlying setting is not exposed to the read APIs this plugin uses (SOQL/Tooling/REST/Metadata read):
280
-
281
- - **Experience Cloud CSP & Lightning Web Security** — the per-site LWR Content Security Policy level and the Lightning Web Security toggle live inside the site's `ExperienceBundle`, not in a `metadata.read`-able type. The check lists live Experience sites and asks for manual verification. A true detection would require retrieving and parsing the `ExperienceBundle` (retrieve → unzip → read `config/*.json`), which is a larger capability than the current read clients; it is tracked as a future enhancement.
282
- - **"Administrators Can Log In as Any User"** is a true detection when SecuritySettings is readable; if the Metadata client is unavailable it degrades to a manual-verification advisory.
283
-
284
- Advisory findings are surfaced as `INFO` and never inflate the health score.
108
+ A handful of checks emit an advisory rather than a verdict, because the underlying setting is not
109
+ exposed to the read APIs this tool uses. Those are listed and explained in the same document —
110
+ advisories are INFO and never inflate the health score.
285
111
 
286
112
  ## Compliance frameworks
287
113
 
288
- Findings are mapped to controls across eight security and privacy frameworks. The mapping is built on a **sourced control catalog** (each control carries its framework, **pinned version**, official title, and a citation) so a finding's compliance reference ties to an exact, defensible requirement rather than a bare tag.
114
+ Findings map to controls across ten frameworks: OWASP Top 10, OWASP LLM Top 10, SOC 2, ISO/IEC
115
+ 27001:2022, the Security Benchmark for Salesforce, the NZ Privacy Act, HISO 10029, NZISM, the HIPAA
116
+ Security Rule, and GDPR.
117
+
118
+ The mapping is built on a **sourced catalog** — each control carries its framework, pinned version,
119
+ official title and a citation — so a finding ties to an exact, defensible requirement rather than a
120
+ bare tag. A **provenance gate** means a control renders only once its reference has been confirmed
121
+ against the authoritative source.
122
+
123
+ Scope the executive report's matrix with `--frameworks universal|nz|all`, or name them:
124
+ `--frameworks owasp,iso,hipaa`. HIPAA and GDPR are in no named pack, because jurisdiction is your
125
+ call, not a default.
289
126
 
290
- | Framework | Version | Notes |
291
- |-----------|---------|-------|
292
- | OWASP Top 10 | 2021 | Web application risk categories |
293
- | OWASP LLM Top 10 | 2025 | LLM/GenAI application risks (LLM01 Prompt Injection, LLM02 Sensitive Information Disclosure, LLM05 Improper Output Handling, LLM06 Excessive Agency); mapped by the AI & Agents checks |
294
- | SOC 2 | AICPA TSC 2017 | Common Criteria (CC6–CC9) |
295
- | ISO/IEC 27001 | 2022 | Annex A controls |
296
- | Security Benchmark for Salesforce (SBS) | current | Salesforce-native benchmark: [docs.securitybenchmark.org](https://docs.securitybenchmark.org) |
297
- | NZ Privacy Act | 2020 | Information Privacy Principles (IPP 5/9/12) |
298
- | HISO 10029 | 2022 | NZ Health Information Security Framework |
299
- | NZISM | v3.8 | NZ Information Security Manual |
127
+ > **Not an attestation.** A control showing "No findings detected" means this audit surfaced no
128
+ > issues mapped to it. It is not a statement of compliance. See [Scope & Liability](#scope--liability).
300
129
 
301
- **Provenance gate.** Each catalogued control is marked `verified` only after its title/reference is confirmed against the authoritative source. **Controls that are not verified do not render** in the compliance matrix. Nothing ships as "compliant-to-clause" on unconfirmed data. The current verification status is tracked in [`docs/compliance/verification-worksheet.md`](docs/compliance/verification-worksheet.md).
130
+ Framework versions, the packs, and the verification status of every control:
131
+ **[docs/COMPLIANCE.md](docs/COMPLIANCE.md)**.
302
132
 
303
- **Framework packs.** The executive report's compliance matrix is scoped with `--frameworks`:
133
+ ## Attack chains
304
134
 
305
- - `universal` *(default)*: OWASP, OWASP LLM Top 10, SOC 2, ISO 27001
306
- - `nz`: ISO 27001, HISO 10029, NZ Privacy Act, NZISM (for NZ health/government engagements)
307
- - `all`: every framework
308
- - a comma list of aliases, e.g. `owasp,owasp-llm,iso,nzism` (`owasp-llm` / `llm` selects the OWASP LLM Top 10)
135
+ A list of findings is not a risk assessment. Three MEDIUM findings that combine into an
136
+ unauthenticated path to bulk data matter more than a lone HIGH that leads nowhere, and reading a
137
+ report severity-by-severity hides exactly that.
309
138
 
310
- > **Not an attestation.** A control rendering "No findings detected" means this audit's checks surfaced no issues mapped to it. It is **not** a statement of compliance or certification. See [Scope & Liability](#scope--liability).
139
+ So every audit correlates its findings into attack chains. **Eleven named chains** are hand-modelled
140
+ scenarios — each with its own narrative and remediation, several naming the concrete request path an
141
+ attacker would use. Where no named chain explains a combination, an emergent pass reports the
142
+ remaining entry-point → outcome pairs as lower-confidence "potential attack paths", so a novel
143
+ combination is still surfaced.
311
144
 
312
- **Further reading:** [Mapping Salesforce security to NZISM, the NZ Privacy Act and ISO 27001](https://softwareinsights.dev/posts/salesforce-security-nzism-nz-privacy-act/) and [Why Salesforce Health Cloud needs its own security review](https://softwareinsights.dev/posts/salesforce-health-cloud-security-review/). More in [Further reading](#further-reading).
145
+ Every chain lists the findings forming its steps, and remediating **any one step breaks the chain**.
146
+ A chain is reported only when every ingredient is present: a clean org produces none.
147
+
148
+ All eleven with severities and trigger conditions: **[docs/ATTACK-CHAINS.md](docs/ATTACK-CHAINS.md)**.
313
149
 
314
150
  ## Scope & Liability
315
151
 
@@ -318,7 +154,7 @@ Findings are mapped to controls across eight security and privacy frameworks. Th
318
154
  **What this tool is not.** It is **not** a penetration test, a dynamic/runtime security test, or a source-code audit of managed-package internals. It does not exploit vulnerabilities, attempt privilege escalation, or guarantee detection of every misconfiguration. The Health Score and A–F grade are **prioritisation aids**, not certifications, and do not represent compliance with, or accreditation under, any standard (OWASP, SOC 2, ISO 27001, HIPAA, GDPR, or otherwise). Compliance-framework tags indicate *relevance* to a control area only.
319
155
 
320
156
 
321
- **Point-in-time.** Results reflect org configuration **at the moment the audit ran**. Configuration drift, new customisations, and platform changes can invalidate findings at any time. Re-run regularly (see [History & Diff](#history--diff)).
157
+ **Point-in-time.** Results reflect org configuration **at the moment the audit ran**. Configuration drift, new customisations, and platform changes can invalidate findings at any time. Re-run regularly (see [History & Diff](docs/COMMANDS.md)).
322
158
 
323
159
  **Authorisation.** Run this tool only against orgs you own or are **explicitly authorised in writing** to assess. You are responsible for obtaining the necessary permissions and for handling generated reports (which may contain sensitive security configuration) in accordance with your organisation's data-handling and confidentiality obligations.
324
160
 
@@ -326,547 +162,91 @@ Findings are mapped to controls across eight security and privacy frameworks. Th
326
162
 
327
163
  ## Trust & verification
328
164
 
329
- Because this tool authenticates against production orgs, "is it safe to run?" deserves a verifiable answer, not just a claim. Here's how you can check for yourself:
330
-
331
- - **Read-only, enforced in CI.** The "no writes to your org" promise is a passing test, not a footnote. `test/unit/invariants/readonly-invariant.test.ts` statically scans this package's entire source tree and fails the build if any jsforce mutation API, HTTP write verb, or bulk/composite write path ever appears. Every org request funnels through the core clients (`@cclabsnz/sf-core`, `src/api/*ClientImpl.ts`), which issue only SOQL queries, REST **GET**s, and Metadata reads. Each package in the monorepo runs the same guard against its own source, so nothing is covered by omission.
332
- - **Nothing phones home, enforced the same way.** `test/unit/invariants/network-egress.test.ts` fails the build on any third-party HTTP client, raw `node:http`/`https` use, telemetry/analytics/LLM endpoint, or websocket — and on any remote asset (`<script src>`, `<link href>`, `@import`) in a generated report. The only network destination is the org you authenticated against. Generated HTML reports are **fully self-contained**: webfonts are embedded as data URIs and Chart.js is inlined, so opening a report never calls out to a CDN — which matters, because reports carry sensitive findings and are often opened offline. Run both guards yourself:
333
-
334
- ```bash
335
- pnpm --filter @cclabsnz/sf-audit test test/unit/invariants
336
- ```
337
- - **What you install matches the public source.** Releases are published from GitHub Actions via [npm trusted publishing (OIDC)](https://docs.npmjs.com/trusted-publishers) with [build provenance](https://docs.npmjs.com/generating-provenance-statements) — no long-lived token, and the npm page shows a signed attestation linking the tarball to the exact public commit that built it. Verify it yourself:
165
+ Because this tool authenticates against production orgs, "is it safe to run?" deserves a verifiable
166
+ answer rather than a claim. Every guarantee below is something you can check yourself:
338
167
 
339
- ```bash
340
- npm audit signatures # reports "verified attestations" for @cclabsnz/sf-audit
341
- ```
168
+ - **Read-only, enforced in CI** — a test statically fails the build if any write path appears in the source
169
+ - **No network egress** — enforced the same way; reports are fully self-contained (fonts and Chart.js inlined)
170
+ - **Signed provenance** — released from CI via npm trusted publishing (OIDC); `npm audit signatures` verifies the tarball against the public commit
171
+ - **Independent scans** — CodeQL (`security-extended`, source *and* workflows), Semgrep, OpenSSF Scorecard, a dependency-review licence gate, `pnpm audit`, Dependabot, secret scanning with push protection, SHA-pinned Actions, and a CycloneDX SBOM per release
342
172
 
343
- - **Independent scans on every change.** Two static-analysis engines — [CodeQL](https://github.com/cclabsnz/sf-audit-plugin/security/code-scanning) (`security-extended`, of both the source **and** the CI workflows) and [Semgrep](https://github.com/cclabsnz/sf-audit-plugin/actions/workflows/semgrep.yml) (OWASP Top 10 + security-audit rulesets) — an [OpenSSF Scorecard](https://securityscorecards.dev/viewer/?uri=github.com/cclabsnz/sf-audit-plugin) supply-chain review, a PR **dependency-review** gate (vulnerabilities **and** a copyleft-license policy), and a `pnpm audit` gate — plus Dependabot, and GitHub secret scanning with push protection. All GitHub Actions are pinned to commit SHAs. Each release ships a CycloneDX **SBOM**.
344
- - **Regulated-environment readiness.** The above give reviewers a paper trail for procurement: SBOM per release, an enforced dependency **license policy**, signed provenance, and independent SAST/supply-chain scans.
345
- - **Least privilege & disclosure.** See [PERMISSIONS.md](PERMISSIONS.md) for the minimal access it needs and [SECURITY.md](SECURITY.md) for private vulnerability reporting.
346
-
347
- ### What third-party scanners flag, and why
348
-
349
- [Socket](https://socket.dev/npm/package/@cclabsnz/sf-audit) raises two **supply-chain risk** alerts against this package. Neither is a vulnerability — both are behavioural heuristics — and rather than suppress them quietly, here is exactly what triggers each and how you can confirm it. The triage is committed as [`socket.yml`](../../socket.yml).
350
-
351
- | Alert | What triggers it | Why it is expected |
352
- | --- | --- | --- |
353
- | **Filesystem access** | `node:fs` reads and writes | It is a CLI that writes your audit reports (HTML/MD/JSON) to disk and reads local inputs: report-branding overrides, event-log baselines under `~/.sf/audit-history`, and its own history archive. Every path is one you pass on the command line or the tool's own dot-directory. |
354
- | **URL strings** | `https://` literals in the shipped code | These are inert citation links rendered as `<a href>` in reports — OWASP, NZISM, the NZ Privacy Act, Te Whatu Ora and CIS-style benchmark references cited by compliance findings. They are never fetched. |
355
-
356
- Check the second one yourself — this lists every URL in the published build:
173
+ Run the guards yourself:
357
174
 
358
175
  ```bash
359
- npm pack @cclabsnz/sf-audit && tar xzf cclabsnz-sf-audit-*.tgz
360
- grep -rhoE 'https?://[a-zA-Z0-9][a-zA-Z0-9.-]*\.[a-zA-Z]{2,}[^"'"'"'`,;) ]*' package/lib | sort -u
176
+ npm test test/unit/invariants
177
+ npm audit signatures
361
178
  ```
362
179
 
363
- As of v1.6.1 that prints seven results, every one a standards-body or documentation link:
364
-
365
- ```
366
- https://docs.securitybenchmark.org/controls-at-a-glance.html
367
- https://genai.owasp.org/llm-top-10/
368
- https://nzism.gcsb.govt.nz/ism-document
369
- https://owasp.org/Top10/2021/
370
- https://privacy.org.nz/privacy-act-2020/privacy-principles/
371
- https://www.legislation.govt.nz/act/public/2020/0031/latest/LMS23342.html
372
- https://www.tewhatuora.govt.nz/health-services-and-programmes/cyber-hub/cyber-standards
373
- ```
374
-
375
- No CDN, telemetry, or analytics endpoints — and the network-egress invariant above fails the build if one is ever added.
180
+ Full detail, including why Socket raises two behavioural alerts against this package and how to
181
+ confirm each: **[docs/TRUST.md](docs/TRUST.md)**. Least-privilege access is in
182
+ [PERMISSIONS.md](PERMISSIONS.md); vulnerability reporting in [SECURITY.md](SECURITY.md).
376
183
 
377
184
  ## Scoring
378
185
 
379
- Each finding is assigned a risk level with a corresponding weight:
380
-
381
- | Risk Level | Default Weight |
382
- |------------|---------------|
383
- | CRITICAL | 10 |
384
- | HIGH | 7 |
385
- | MEDIUM | 4 |
386
- | LOW | 1 |
387
- | INFO | 0 |
388
-
389
- The health score is calculated as `100 - (total weight / max possible weight) * 100`, capped at 0.
390
-
391
- The audit produces a **Health Score** (0–100) and a **Grade** (A–F):
392
-
393
- | Grade | Criteria |
394
- |-------|---------|
395
- | A | Score ≥ 85, no HIGH findings |
396
- | B | Score ≥ 70, ≤ 1 HIGH finding |
397
- | C | Score ≥ 55, ≤ 3 HIGH findings |
398
- | D | Score ≥ 40, no CRITICAL findings |
399
- | F | Score < 40 or any CRITICAL finding |
400
-
401
- ### Custom Scoring Config
402
-
403
- All weights and grade thresholds are configurable: no recompile needed. This is useful when your org has a different risk appetite (e.g. you want to penalise hardcoded credentials more heavily, or set stricter grade thresholds).
404
-
405
- **Step 1:** Copy the sample config as your starting point:
406
-
407
- ```bash
408
- cp config/scoring.sample.json my-scoring.json
409
- ```
410
-
411
- **Step 2:** Edit the values. All three sections (`riskScores`, `checkWeights`, `gradeThresholds`) are optional: omit any section to keep the defaults.
412
-
413
- ```json
414
- {
415
- "riskScores": {
416
- "CRITICAL": 10,
417
- "HIGH": 7,
418
- "MEDIUM": 4,
419
- "LOW": 1,
420
- "INFO": 0
421
- },
422
- "checkWeights": {
423
- "hardcoded-credentials": 10,
424
- "guest-user-access": 10,
425
- "users-and-admins": 10,
426
- "apex-sharing": 7
427
- },
428
- "gradeThresholds": {
429
- "A": { "minScore": 90, "maxHigh": 0 },
430
- "B": { "minScore": 75, "maxHigh": 1 },
431
- "C": { "minScore": 60, "maxHigh": 3 },
432
- "D": { "minScore": 40, "maxCritical": 0 },
433
- "F": {}
434
- }
435
- }
436
- ```
437
-
438
- The full list of valid `checkWeights` keys (one per check) is in [`config/scoring.sample.json`](config/scoring.sample.json).
439
-
440
- **Step 3:** Pass it when running the audit:
441
-
442
- ```bash
443
- sf audit security --target-org myOrg --scoring-config ./my-scoring.json
444
- ```
445
-
446
- Your config is deep-merged with the defaults, so you only need to include the values you want to change.
447
-
448
- ## History & Diff
449
-
450
- Every `sf audit security` run automatically archives a JSON copy of the report to:
451
-
452
- ```
453
- ~/.sf/audit-history/{orgId}/sf-audit-{orgId}-{timestamp}.json
454
- ```
455
-
456
- No configuration needed: archiving happens silently after each run.
457
-
458
- ### View Audit History
459
-
460
- Show how your org's security posture has changed across multiple runs:
461
-
462
- ```bash
463
- sf audit history --target-org myOrg
464
- ```
465
-
466
- Prints a terminal table with score trends and writes an HTML timeline to the current directory.
467
-
468
- **Flags:**
469
-
470
- | Flag | Description | Default |
471
- |------|-------------|---------|
472
- | `--target-org` | Org alias or username | required |
473
- | `--reports-dir` | Custom directory containing archived reports | `~/.sf/audit-history/{orgId}` |
474
- | `--output` | Directory to write the HTML timeline | `.` (cwd) |
475
- | `--limit` | Maximum number of most-recent runs to show | all |
186
+ Each finding carries a risk weight (CRITICAL 10, HIGH 7, MEDIUM 4, LOW 1, INFO 0). The health score is
187
+ `100 - (total weight / max possible weight) * 100`, capped at 0, and maps to an A–F grade — A needs
188
+ ≥ 85 with no HIGH findings; any CRITICAL is an F.
476
189
 
477
- **Example output:**
190
+ All weights and grade thresholds are configurable without recompiling, via
191
+ `--scoring-config ./my-scoring.json`. Start from [`config/scoring.sample.json`](config/scoring.sample.json),
192
+ which lists every valid `checkWeights` key; your config is deep-merged with the defaults, so include
193
+ only what you want to change.
478
194
 
479
- ```
480
- Audit History: My Org (00D000000000001)
481
- ────────────────────────────────────────────────────────────────────────────────
482
- # Date Score Grade CRIT HIGH MED LOW Δ Score
483
- ────────────────────────────────────────────────────────────────────────────────
484
- 1 2026-03-23 15:10 64 D 1 5 8 3 —
485
- 2 2026-04-09 11:22 81 B 0 2 5 3 +17
486
- ────────────────────────────────────────────────────────────────────────────────
487
- Trend: ▲ +17 over 2 audits Best: 81 (2026-04-09 11:22) Worst: 64 (2026-03-23 15:10)
488
- ```
489
-
490
- ### Diff Two Reports
491
-
492
- Compare any two audit JSON files to see exactly what changed:
493
-
494
- ```bash
495
- sf audit diff baseline.json current.json
496
- ```
497
-
498
- Writes an HTML and JSON diff report to the current directory.
499
-
500
- **Flags:**
195
+ Grade bands, the full config shape and worked examples: **[docs/SCORING.md](docs/SCORING.md)**.
501
196
 
502
- | Flag | Description | Default |
503
- |------|-------------|---------|
504
- | `--output` | Directory to write diff reports | `.` (cwd) |
505
- | `--format` | Comma-separated formats: `html`, `json` | `html,json` |
197
+ ## Other commands
506
198
 
507
- **Example output:**
508
-
509
- ```
510
- Diff report written: ./sf-audit-diff-00D000000000001-...-vs-....html
511
- Diff report written: ./sf-audit-diff-00D000000000001-...-vs-....json
512
-
513
- ─────────────────────────────
514
- Diff Summary
515
- ─────────────────────────────
516
- Score delta +17
517
- Grade D → B
518
- New 0
519
- Resolved 1
520
- ─────────────────────────────
521
- ```
199
+ Beyond the audit itself, the plugin ships four commands. Each is documented in
200
+ **[docs/COMMANDS.md](docs/COMMANDS.md)**.
522
201
 
523
- ## Free event baseline
202
+ | Command | What it does |
203
+ |---------|--------------|
204
+ | `sf audit history` / `sf audit diff` | Every run auto-archives to `~/.sf/audit-history/{orgId}`. Show posture drift across runs as a table plus an HTML timeline, or diff any two report JSONs |
205
+ | `sf audit events pull` | Capture the org's **free** daily `EventLogFile` logs to local disk before the ~1-day retention window drops them — no Event Monitoring / Shield add-on needed. Idempotent, safe to cron |
206
+ | `sf audit timeline` | Reconstruct one actor's activity across every captured event type, entirely offline. Refuses to expand a shared identity or join on a blank field, and always reports capture coverage first |
207
+ | `sf audit apps` | Read the `RestApi` event log to compare what each connected app actually *uses* against what its run-as user is *granted*, and emit a suggested least-privilege permission set |
524
208
 
525
- Salesforce's free tier exposes **Daily-interval `EventLogFile` logs** (login, API, and error
526
- activity) on Enterprise/Unlimited/Performance editions and Developer Edition — *without* the paid
527
- Event Monitoring / Shield add-on. The catch: on the free tier those logs are retained for only
528
- **~1 day**. Miss a day and that day's activity is gone.
529
-
530
- `sf audit events pull` captures them to local disk before they expire, so a daily run builds a
531
- rolling local baseline you own:
532
-
533
- ```bash
534
- sf audit events pull --target-org myOrg
535
- ```
536
-
537
- It queries whatever daily event types the org actually exposes, downloads each log's CSV body, and
538
- saves it to `~/.sf/event-baseline/{orgId}/{EventType}/{LogDate}-{Id}.csv`, plus a per-run manifest.
539
- It is **read-only** (GET only) and **idempotent**: any log already on disk is skipped, so it is safe
540
- to run repeatedly. Run it once a day from cron or a scheduled GitHub Action and you beat the 1-day
541
- retention window with a growing archive — no add-on required.
542
-
543
- ```bash
544
- # Daily cron entry (07:15) — capture yesterday's logs
545
- 15 7 * * * sf audit events pull --target-org myOrg >> ~/.sf/event-baseline/pull.log 2>&1
546
- ```
547
-
548
- **Flags:**
549
-
550
- | Flag | Description | Default |
551
- |------|-------------|---------|
552
- | `--target-org` | Org alias or username | required |
553
- | `--since` | Days of `LogDate` to request (`LAST_N_DAYS` window) | `1` |
554
- | `--types` | Restrict to specific EventTypes, comma-separated (e.g. `Login,ApiTotalUsage`) | *(all available)* |
555
- | `--output` / `-o` | Base directory to store logs under | `~/.sf/event-baseline` |
556
-
557
- ```bash
558
- # Backfill the last 3 days (each still subject to the org's retention)
559
- sf audit events pull --target-org myOrg --since 3
560
-
561
- # Only pull login and API-usage logs
562
- sf audit events pull --target-org myOrg --types Login,ApiTotalUsage
563
-
564
- # Store under a project-local directory instead of ~/.sf
565
- sf audit events pull --target-org myOrg --output ./event-baseline
566
- ```
567
-
568
- > Reading `EventLogFile` requires the **View Event Log Files** permission on the running user (this
569
- > is in addition to the minimum read-only audit permission set). If it is missing, or the edition
570
- > does not expose free daily logs, the command exits cleanly with an explanation rather than failing.
571
-
572
- **Example output:**
573
-
574
- ```
575
- Pulling free EventLogFile logs for org: My Org (00D000000000001)
576
-
577
- ─────────────────────────────
578
- Event Baseline Pull
579
- ─────────────────────────────
580
- Found 7
581
- Downloaded 7
582
- Skipped 0 (already saved)
583
- Total bytes 48213
584
- ─────────────────────────────
585
- Saved to: ~/.sf/event-baseline/00D000000000001
586
- Manifest: ~/.sf/event-baseline/00D000000000001/_manifests/manifest-...-....json
587
- ```
588
-
589
- ### Analyzing the captured logs
590
-
591
- `events pull` is the collection half. To triage those `EventLogFile` CSVs for exploit and
592
- abuse patterns, use the companion CLI **[sfelf-triage](https://github.com/cclabsnz/sfelf-triage)**.
593
- It reads downloaded EventLogFile CSVs and emits a per-IP verdict
594
- (`BENIGN_SCANNER | SUSPICIOUS | LIKELY_ABUSE`), answering *"is this guest/community IP a
595
- vulnerability scanner or a real threat?"* — with **zero network egress** and no org connection.
596
-
597
- sfelf-triage reads this plugin's `~/.sf/event-baseline/<orgId>` layout directly, so the two
598
- tools chain with no glue:
599
-
600
- ```bash
601
- sf audit events pull --target-org myOrg # capture (this plugin)
602
- sfelf-triage analyze ~/.sf/event-baseline/<orgId> # triage (companion)
603
- ```
604
-
605
- See the [sfelf-triage README](https://github.com/cclabsnz/sfelf-triage#readme) for install and usage.
606
-
607
- ## Forensic timeline
608
-
609
- Salesforce splits one actor's activity across many event types, and each carries a different
610
- subset of identifying fields — so no single log answers *"what did this actor do"*. Filtering by
611
- IP finds the requests but misses every SOQL execution, because the query log has no `CLIENT_IP`
612
- column at all. Filtering by user finds everything the *guest* user did, which on a community is
613
- everyone.
614
-
615
- `sf audit timeline` correlates a seed across every captured event type and writes a defensible
616
- timeline:
617
-
618
- ```bash
619
- sf audit timeline --window 2026-08-02T04:00Z/PT1H --seed ip:203.0.113.50
620
- ```
621
-
622
- It runs **entirely offline** against captures written by `sf audit events pull` — no org
623
- connection is opened, so it still works long after the org's retention window has expired or its
624
- credentials have been revoked. The org id is inferred from the capture directory when only one
625
- org has been captured.
626
-
627
- ### What it will not do
628
-
629
- Cross-event correlation is easy to get confidently wrong, and a wrong answer here reads as
630
- evidence. Two rules are enforced and both are visible in the output:
631
-
632
- - **A blank field is never a join key.** A blank `REQUEST_ID` used as a key stops identifying
633
- anything and starts matching every other row whose value is also blank — quietly attributing
634
- strangers' sessions to your actor.
635
- - **A shared identity is not expanded.** A community guest user can stand for hundreds of
636
- distinct visitors. Expanding through it would present the whole crowd's activity as one
637
- actor's, so it is refused by default and the refusal is reported with its evidence:
638
-
639
- ```
640
- Expansion refused: userId 005xx0000000000 is shared by 1371 distinct addresses
641
- (threshold 8). Override with --allow-shared-identity.
642
- ```
643
-
644
- Every output also leads with what was actually captured, because *"no rows matched"* means two
645
- entirely different things — the actor did nothing, or nobody captured the hour they did it in:
646
-
647
- ```
648
- Window — coverage INCOMPLETE
649
- captured AuraRequest, ListViewEvent
650
- MISSING LightningInteraction (not-in-core-set)
651
- MISSING GuestUserAnomalyEventStore (storage-disabled)
652
-
653
- No activity in captured sources. Coverage incomplete — 2 sources missing.
654
- ```
655
-
656
- **Flags:**
657
-
658
- | Flag | Description | Default |
659
- |------|-------------|---------|
660
- | `--window` | When to look — see below | required |
661
- | `--seed` | Typed and repeatable: `ip:` `user:` `session:` `request:` `login:` `transaction:` `event:` | *(whole window)* |
662
- | `--org-id` | Which captured org to read | *(inferred when unambiguous)* |
663
- | `--input` | Capture base directory | `~/.sf/event-baseline` |
664
- | `--allow-shared-identity` | Expand through identities shared by many actors | `false` |
665
- | `--max-cardinality` | Distinct-actor ceiling above which a key is not expanded | `8` |
666
- | `--format` | Comma-separated: `csv,json,md` | all three |
667
- | `--output` | Directory to write into | `.` |
668
-
669
- **Finding your way in.** Two things you do not have to know up front. Omit `--seed` and you get
670
- the whole window, uncorrelated — which is where you look to find something worth seeding on.
671
- Ask for a window that was never captured and the error lists the days that *were*:
672
-
673
- ```
674
- No captures for 2026-07-01 under ~/.sf/event-baseline/00Dxx0000000000EAA.
675
-
676
- Captured days for this org:
677
- 2026-08-01 11 event type(s), whole day
678
- 2026-08-02 14 event type(s), whole day
679
-
680
- Try: --window 2026-08-02
681
- ```
682
-
683
- **Saying when.** `--window` takes whichever form is nearest to hand — you should not have to
684
- compose an ISO 8601 interval while an incident is running:
685
-
686
- | You type | You get |
687
- |---|---|
688
- | `yesterday` | the whole of yesterday, UTC |
689
- | `today` | midnight UTC until now |
690
- | `2h` · `90m` | the last two hours; the last ninety minutes |
691
- | `2026-08-02` | that whole day — the shape the free tier captures in |
692
- | `2026-08-02T04:17Z` | the hour containing that instant, for a timestamp pasted from an alert |
693
- | `2026-08-02T04:00Z/PT1H` | an exact interval, start and duration |
694
- | `2026-08-02T04:00Z/2026-08-02T06:00Z` | an exact interval, start and end |
695
-
696
- Times are UTC, because every capture is stored in UTC — a bare timestamp with no zone is read
697
- that way rather than as local time. A window has to fall inside one UTC day; one that crosses
698
- midnight is refused, and the error names the two runs that would cover it.
699
-
700
- ```bash
701
- # Follow one address across every captured event type
702
- sf audit timeline --window yesterday --seed ip:203.0.113.50
703
-
704
- # Start from a request and walk outward, including its Apex cascade
705
- sf audit timeline --window 2026-08-02T04:00Z/PT1H --seed request:abc123
706
-
707
- # Several orgs captured — name the one to read
708
- sf audit timeline --org-id 00Dxx0000000000EAA --window 2026-08-02T04:00Z/PT2H --seed user:005xx000000000
709
-
710
- # Machine-readable only, into an evidence directory
711
- sf audit timeline --window 2026-08-02T04:00Z/PT1H --seed ip:203.0.113.50 \
712
- --format json --output ./evidence/incident-2026-08-02
713
- ```
714
-
715
- **Outputs:**
716
-
717
- | File | Contents |
718
- |------|----------|
719
- | `timeline.csv` | The correlated rows, one schema across all sources, chronological |
720
- | `timeline.json` | The same rows plus the provenance — seeds, every key expanded through, and every refusal |
721
- | `summary.md` | Narrative: coverage, per-type counts, what tied each row in, refusals, and whether records left |
722
-
723
- Each row records **which join key tied it in**, so attribution is checkable rather than asserted.
724
- The `rows_processed` and `records_returned` columns are populated only from Real-Time Event
725
- objects — no `EventLogFile` type records them — and when none were captured the summary says the
726
- question is unanswerable rather than going quiet, since silence there reads as *"nothing left"*.
727
-
728
- ### Checking a claim against its control group
729
-
730
- When a seed's rows share an identity with many other people — a community guest user, a shared
731
- integration account — *"this actor did X"* needs something to be checked against. Otherwise the
732
- claim is unfalsifiable: nobody can tell your 15 rows from the other 600 behind the same user.
733
-
734
- The refusal message already gives you the denominator:
735
-
736
- ```
737
- Expansion refused: userId 005xx000000000 is shared by 1371 distinct addresses (threshold 8).
738
- ```
739
-
740
- To see the peer set itself, run the command a second time seeded on that identity, with expansion
741
- allowed:
742
-
743
- ```bash
744
- # 1. The claim — narrow, seeded on something that identifies one actor
745
- sf audit timeline --window 2026-08-02T04:00Z/PT1H --seed request:REQ000000 \
746
- --output ./evidence/actor
747
-
748
- # 2. The control — everyone behind the identity those rows share
749
- sf audit timeline --window 2026-08-02T04:00Z/PT1H --seed user:005xx000000000 \
750
- --allow-shared-identity --output ./evidence/peers
751
- ```
752
-
753
- Both use the same schema, so they concatenate and diff directly:
754
-
755
- ```bash
756
- # How much of the identity's activity is actually your actor?
757
- # (tail -n +2 skips the header row, so these are data rows)
758
- echo "actor: $(tail -n +2 ./evidence/actor/timeline.csv | wc -l)"
759
- echo "control: $(tail -n +2 ./evidence/peers/timeline.csv | wc -l)"
760
- ```
761
-
762
- An actor accounting for 15 of 603 rows — 2.5%, and separable by request id — is a different
763
- finding from one accounting for 580 of 603. The second command is what lets a reviewer tell
764
- which they are looking at, rather than taking the first on trust.
765
-
766
- > `sf audit timeline` adds **no checks** and does not affect the security grade. It is an
767
- > investigation command, not a `SecurityCheck`.
768
-
769
- ## Connected-app least-privilege
770
-
771
- Connected-app over-privilege was at the centre of the 2025-2026 wave of Salesforce data-theft
772
- via OAuth: apps authorized with more scope than they use, and integration users with far more
773
- object access than the app ever touches. The static checks (`connected-apps`, `connected-app-scope`,
774
- `connected-app-inactivity`) tell you what was *granted*. `sf audit apps` tells you what is actually
775
- *used*, so it can point at the specific access to remove.
776
-
777
- ```bash
778
- sf audit apps --target-org myOrg --since 7
779
- ```
780
-
781
- It reads the `RestApi` `EventLogFile` to see which objects each connected app touched, compares that
782
- against the objects its run-as user can reach, and reports the over-grant per object and read/write
783
- bit — plus a generated least-privilege permission set granting exactly what was observed. App IDs are
784
- resolved to human-readable names (`AppMenuItem` / `ConnectedApplication` / a bundled standard-app
785
- catalog / `LoginHistory` correlation), and anything unresolved is flagged loudly rather than hidden.
786
-
787
- It is **read-only**: the suggested permission set is emitted as data, never deployed.
788
-
789
- **Honest bounds.** `RestApi` attributes roughly half of API traffic to a connected app (the rest is
790
- UI-API / session traffic), so *used* is a lower bound. Findings carry the observation window and
791
- attribution rate, and revoke recommendations are suppressed below a soak window and for apps used by
792
- many interactive users. Reading `EventLogFile` needs the **View Event Log Files** permission (the same
793
- one `events pull` uses).
794
-
795
- **Flags:**
796
-
797
- | Flag | Description | Default |
798
- |------|-------------|---------|
799
- | `--target-org` | Org alias or username | required |
800
- | `--since` | Days of `RestApi` log to analyze | `7` |
801
- | `--from` | Read `RestApi` CSVs from a local `events pull` baseline dir instead of downloading | *(download)* |
802
- | `--soak` | Minimum window (days) before asserting revoke recommendations | `7` |
803
- | `--format` | `table` / `json` / `md` | `table` |
804
-
805
- ```bash
806
- # Reuse an events-pull baseline instead of downloading again
807
- sf audit apps --target-org myOrg --from ~/.sf/event-baseline/00Dxxx
808
-
809
- # Machine-readable output
810
- sf audit apps --target-org myOrg --format json
811
- ```
209
+ To triage captured logs for abuse patterns, pair `events pull` with the companion CLI
210
+ **[sfelf-triage](https://github.com/cclabsnz/sfelf-triage)**, which reads this plugin's
211
+ `~/.sf/event-baseline/<orgId>` layout directly.
812
212
 
813
213
  ## Requirements
814
214
 
815
215
  - Node.js 18+
816
216
  - Salesforce CLI (`sf`) v2+
817
- - A least-privilege, **read-only** org user. The audit performs no writes and does **not** require `View All Data`. See **[PERMISSIONS.md](PERMISSIONS.md)** for the exact minimum permission set, what each is for, what the tool does *not* need, and a ready-to-deploy `SF Audit (Read-Only)` permission set ([`docs/permissionset/`](docs/permissionset/SF_Audit_ReadOnly.permissionset-meta.xml)).
217
+ - A least-privilege, **read-only** org user. The audit performs no writes and does **not** require `View All Data` — without it one probe (integration write evidence) reports *inconclusive* rather than drawing a conclusion. See **[PERMISSIONS.md](PERMISSIONS.md)** for the exact minimum permission set, what each is for, what the tool does *not* need, and a ready-to-deploy `SF Audit (Read-Only)` permission set ([`docs/permissionset/`](docs/permissionset/SF_Audit_ReadOnly.permissionset-meta.xml)).
818
218
 
819
219
  ## Development
820
220
 
821
221
  ```bash
822
222
  npm run build # compile TypeScript
823
- npm test # run all tests
824
- npm run test:unit # unit tests only
223
+ npm run typecheck # type-check src + test (tsc --noEmit)
224
+ npm test # typecheck, then run all tests
225
+ npm run test:unit # typecheck, then unit tests only
226
+ npm run test:jest # tests without the typecheck — fast inner loop
825
227
  npm run clean # remove compiled output
826
228
  ```
827
229
 
828
- Maintainers: see **[docs/RELEASE.md](docs/RELEASE.md)** for the release checklist and the one-time repository-hardening steps (npm provenance token, branch protection, and the CodeQL / Scorecard setup behind the badges above).
829
-
830
- ## Further reading
230
+ Jest transforms with [swc](https://swc.rs), which strips types without checking them, so
231
+ `npm run typecheck` is what catches type errors in tests — it runs ahead of Jest in
232
+ `npm test`. Use `npm run test:jest` while iterating, but don't treat it as a green run.
831
233
 
832
- Deep dives on this tool and the topics it checks, from our engineering blog **[softwareinsights.dev](https://softwareinsights.dev)**.
234
+ Release history is in **[CHANGELOG.md](CHANGELOG.md)**; each entry mirrors its
235
+ [GitHub Release](https://github.com/cclabsnz/sf-audit-plugin/releases), which carries the signed
236
+ provenance attestation and the CycloneDX SBOM for that build.
833
237
 
834
- **How the commands work:**
238
+ Project documents: **[GOVERNANCE.md](docs/GOVERNANCE.md)** (roles, decisions, continuity), **[ARCHITECTURE.md](docs/ARCHITECTURE.md)**, **[ROADMAP.md](docs/ROADMAP.md)**, and the **[assurance case](docs/ASSURANCE-CASE.md)** — the argument, with evidence, that this tool is safe to point at a production org.
835
239
 
836
- - [How sf-audit works — checks, attack chains, and compliance mapping](https://softwareinsights.dev/posts/sf-audit-61-checks-attack-chains-compliance-mapping/) — the design walkthrough behind `sf audit security`
837
- - [Free Salesforce Event Monitoring: a baseline from EventLogFile without Shield](https://softwareinsights.dev/posts/salesforce-free-event-monitoring-eventlogfile-baseline/) — why [`sf audit events pull`](#free-event-baseline) exists and how to cron it
838
- - [Which connected apps use less than they're granted? Ask your own logs](https://softwareinsights.dev/posts/salesforce-connected-app-least-privilege-granted-vs-used/) — the granted-vs-used method behind [`sf audit apps`](#connected-app-least-privilege)
839
- - [Scanner or breach? Triage EventLogFile in one command](https://softwareinsights.dev/posts/salesforce-eventlogfile-guest-traffic-triage-scanner-or-breach/) — pairing an `events pull` baseline with the [sfelf-triage](https://github.com/cclabsnz/sfelf-triage) companion
840
- - [Salesforce guest user exposure, graded by real reachability](https://softwareinsights.dev/posts/sf-audit-guest-user-exposure-reachability/) — the UI-API reachability tiering behind the [guest checks](#guest--external-facing-access)
841
- - [sf-audit vs sf-cli-security-audit](https://softwareinsights.dev/posts/sf-audit-vs-sf-cli-security-audit/) — how this plugin differs from a configurable policy engine, and when to reach for each
842
-
843
- **The access model the privilege checks encode** — the reasoning behind [Users, Permissions & Privilege](#users-permissions--privilege):
844
-
845
- - [Why your developers don't need Modify All Data](https://softwareinsights.dev/posts/salesforce-developers-modify-all-data-what-they-need-instead/) — part 1 of a four-part series on delivery-team access; feeds Users & Admins and Privileged Access & Shadow Admins
846
- - [The access model: tiers and roles](https://softwareinsights.dev/posts/salesforce-delivery-team-access-model-tiers-and-roles/) — the tiering that Separation of Duties and Privilege Escalation Permissions test against
847
- - [Sizing the model to your team](https://softwareinsights.dev/posts/salesforce-access-model-sizing-internal-vs-external-admin-teams/) — internal vs external admins, and what the model costs to run
848
- - [Deployable permission sets](https://softwareinsights.dev/posts/salesforce-delivery-team-deployable-permission-sets/) — the metadata and the CI identity, which the Integration / Service Accounts check inventories
849
-
850
- **Platform changes the checks track:**
240
+ Maintainers: see **[docs/RELEASE.md](docs/RELEASE.md)** for the release checklist and the one-time repository-hardening steps (npm provenance token, branch protection, and the CodeQL / Scorecard setup behind the badges above).
851
241
 
852
- - [Hardening Agentforce against prompt injection (post-ForcedLeak)](https://softwareinsights.dev/posts/salesforce-agentforce-forcedleak-prompt-injection-hardening/) — feeds the Agentforce / GenAI checks
853
- - [Audit your Agentforce footprint: every agent, agent user, and permission](https://softwareinsights.dev/posts/salesforce-agentforce-footprint-audit/) — the manual SOQL behind Agent Inventory
854
- - [Agentforce agent user least privilege](https://softwareinsights.dev/posts/salesforce-agentforce-agent-user-least-privilege/) — feeds the Agent User Privilege check
855
- - [Salesforce MFA enforcement: the revised 2026 dates](https://softwareinsights.dev/posts/salesforce-mfa-enforcement-paused-revised-dates-2026/) — feeds the MFA checks
856
- - [The MFA enforcement admin guide](https://softwareinsights.dev/posts/salesforce-mfa-enforcement-2026-admin-guide/) — what the MFA Enforcement / Registration / Method Strength checks are measuring against
857
- - [MFA lockout recovery and break-glass accounts](https://softwareinsights.dev/posts/salesforce-mfa-lockout-recovery-break-glass-accounts/) — the operational side of the MFA and High Assurance Session checks
858
- - [OAuth username-password (ROPC) flow retirement in Winter '27](https://softwareinsights.dev/posts/salesforce-oauth-username-password-flow-retirement-winter-27/) — feeds the SSO Enforcement and Connected App OAuth Scopes checks
859
- - [Email change verification retirement and Authorized Email Domains](https://softwareinsights.dev/posts/salesforce-email-change-verification-retirement-authorized-email-domains/) — feeds the Email Security check
860
- - [Summer '26: SAML retirement & Apex secure-by-default](https://softwareinsights.dev/posts/salesforce-summer-26-saml-retirement-apex-secure-by-default/) — feeds the SSO / Apex checks
861
- - [Salesforce security enforcement in 2026 — every change and date](https://softwareinsights.dev/posts/salesforce-security-enforcement-2026-complete-guide/) — the overall posture this tool measures
862
- - [Report-export step-up enforcement: known issues](https://softwareinsights.dev/posts/salesforce-transaction-security-policy-report-export-known-issues/) — feeds the data-export / transaction-security checks
242
+ ## Further reading
863
243
 
864
- **Compliance and NZ context** — background for the [framework mappings](#compliance-frameworks):
244
+ Deep dives on this tool and the topics it checks, from our engineering blog
245
+ **[softwareinsights.dev](https://www.softwareinsights.dev)** — including how sf-audit works, the free
246
+ Event Monitoring baseline, guest-user exposure grading, the delivery-team access model, Agentforce
247
+ hardening after ForcedLeak, and the NZ compliance context.
865
248
 
866
- - [Mapping Salesforce security to NZISM, the NZ Privacy Act and ISO 27001](https://softwareinsights.dev/posts/salesforce-security-nzism-nz-privacy-act/)
867
- - [Data sovereignty for New Zealand Salesforce orgs](https://softwareinsights.dev/posts/salesforce-data-sovereignty-new-zealand/) — residency vs sovereignty, and why the `nz` framework pack exists
868
- - [IPP 3A indirect collection notices](https://softwareinsights.dev/posts/salesforce-nz-privacy-ipp3a-indirect-collection-notice/) — feeds the NZ Privacy Act control mappings
869
- - [Why Salesforce Health Cloud needs its own security review](https://softwareinsights.dev/posts/salesforce-health-cloud-security-review/) — the HISO 10029 context
249
+ Full annotated list: **[docs/FURTHER-READING.md](docs/FURTHER-READING.md)**.
870
250
 
871
251
  ## Commercial support
872
252