Audits SCC harness configuration in Salesforce projects for completeness, scoring hooks, agents, skills, and rules coverage for Apex and LWC development.
How this skill is triggered — by the user, by Claude, or both
Slash command
/salesforce-claude-code:sf-harness-auditThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Evaluate the SCC harness configuration and suggest improvements. Checks file existence, content coverage, hook status, and rule completeness.
Evaluate the SCC harness configuration and suggest improvements. Checks file existence, content coverage, hook status, and rule completeness.
Check what's installed in the current project:
# Check for SCC installation markers
ls -la .claude/hooks/scripts/ 2>/dev/null
ls -la .claude/agents/ 2>/dev/null
ls -la .claude/skills/ 2>/dev/null
If using SCC CLI: npx scc-universal status
Rate each category 0-10 using these rubrics:
Hook Coverage (0-10)
| Score | Criteria |
|---|---|
| 0-3 | No hooks or only SessionStart |
| 4-6 | Quality-gate and governor-check active, but missing pre-tool-use hooks |
| 7-8 | All standard profile hooks active |
| 9-10 | Strict profile enabled, all hooks active including Code Analyzer integration |
Check: npx scc-universal status to see installed hooks and their profiles.
Check: echo $SCC_HOOK_PROFILE to verify profile level.
Agent Coverage (0-10)
| Score | Criteria |
|---|---|
| 0-3 | Only platform agents (loop-operator, doc-updater) |
| 4-6 | Core domain agents (sf-architect, sf-apex-agent, sf-review-agent) |
| 7-8 | Full domain agents (+ sf-lwc-agent, sf-flow-agent, sf-integration-agent) |
| 9-10 | Complete coverage including sf-admin-agent, sf-agentforce-agent, sf-bugfix-agent |
Check: ls .claude/agents/sf-*.md 2>/dev/null | wc -l
Skill Coverage (0-10)
| Score | Criteria |
|---|---|
| 0-3 | Less than 10 skills |
| 4-6 | 10-20 skills, core SF patterns covered |
| 7-8 | 20-35 skills, enterprise patterns included |
| 9-10 | 35+ skills with full domain coverage |
Check: find .claude/skills/ -name "SKILL.md" | wc -l
Skill Coverage (User-Invocable) (0-10)
| Score | Criteria |
|---|---|
| 0-3 | Only basic skills (sf-architect agent, sf-apex-best-practices) |
| 4-6 | Core workflow skills (sf-deployment, sf-debugging, sf-security) |
| 7-8 | Testing and security skills (sf-tdd-workflow, sf-governor-limits) |
| 9-10 | Full suite including discovery (/sf-help, /sf-quickstart) |
Check: ls .claude/skills/sf-*/SKILL.md 2>/dev/null | wc -l
Security Posture (0-10)
| Score | Criteria |
|---|---|
| 0-3 | No security hooks or rules |
| 4-6 | Security rules present, quality-gate active |
| 7-8 | Governor-check active, sharing model detection, CRUD/FLS rules |
| 9-10 | Strict profile, Code Analyzer integration, security-reviewer agent active |
SCC Harness Audit
══════════════════════════════════════════
Profile: standard
Version: 1.0.0
Install Target: claude
Category Scores:
Hook Coverage: 7/10
Agent Coverage: 8/10
Rule Coverage: 6/10
Skill Coverage: 7/10
Skill (Invocable): 8/10
Security Posture: 7/10
────────────────────────
Overall Score: 7.2/10
Top Recommendations:
1. [+1.0] Enable strict hook profile: export SCC_HOOK_PROFILE=strict
2. [+0.5] Install all modules: npx scc-universal install all
For each gap, provide a specific command to fix it:
| Gap | Fix Command |
|---|---|
| Missing hooks | npx scc-universal repair |
| Missing domain rules | npx scc-universal install all (or target: npx scc-universal install apex, npx scc-universal install lwc) |
| Low skill count | npx scc-universal install all |
| Wrong hook profile | export SCC_HOOK_PROFILE=strict |
| Drifted files | npx scc-universal doctor then npx scc-universal repair |
sf-harness-audit
sf-harness-audit Check if all Salesforce domains are covered by agents and skills
sf-harness-audit Report gaps in hook coverage and recommend improvements
sf-harness-audit Score the security posture of the current SCC installation
/sf-harness-audit — This skill is the audit tool; no separate agent existsnpx claudepluginhub jiten-singh-shahi/salesforce-claude-code --plugin salesforce-claude-codeDiagnoses Claude Code harness health (hooks, skills, agents, rules, MCP, eval) across 8 dimensions, scores 0-24 with S-D grades, and provides improvement suggestions. Scans ~/.claude/. Triggers: harness audit, 하네스 진단.
Reviews Salesforce DevSecOps pipelines: SCA finding triage, sandbox data masking, change impact analysis, CI/CD security gates, and DevOps Center governance. Works from sanitized exports only.
Runs Salesforce Code Analyzer to scan Apex, LWC, and JS code for security, performance, style, and duplicate violations. Supports all engines (PMD, ESLint, CPD, RetireJS, Flow, SFGE, ApexGuru) and targets including git diff.