Second Brain

claude-code-to-salesforce-expert-20260426-org-health-followup

2 min read

Salesforce Expert — P1.1 + P2.1 + P2.2 Followup Investigation

Context

Mon health pull (2026-04-26) flagged 3 P1 + 2 P2 anomalies. P1.2 + P1.3 are storage-strategy decisions awaiting AJ. The remaining three are autonomous-actionable expert-scope work.

Tasks (next heartbeat cycle)

Task 1 — P1.1 Identify offending Scheduled Path Flow (READ-ONLY investigation)

Diagnostic only — no remediation in this cycle, just naming the culprit Flow + estimating run rate.

  1. sf_query SELECT Id, MasterLabel, Status, ProcessType, TriggerType, RunInMode FROM FlowDefinitionView WHERE TriggerType IN ('RecordAfterSave','RecordBeforeSave') AND Status='Active'
  2. For each candidate, sf_get_flow_execution_logs over last 24h to count Scheduled Path firings
  3. Rank by 24h firing count desc; top 3 named in followup vault entry
  4. Hypothesize selectivity gap (entry criteria too broad, schedule too tight, or high-volume object)

Acceptance: vault entry salesforce-p1.1-flow-investigation-2026-04-26.md with named Flow + estimated daily run count + selectivity hypothesis. AJ decides remediation path next.

Task 2 — P2.1 Inactive User Reclaim (REVERSIBLE — autonomous)

Per expert-recommendations rule: deactivation is reversible (reactivate any time), within scope.

  1. sf_query SELECT Id, Name, Username, LastLoginDate, IsActive FROM User WHERE IsActive=true AND LastLoginDate < LAST_N_DAYS:90
  2. Cross-check against any “do-not-deactivate” list (executives, integration users, scheduled-job owners) — exclude those
  3. Generate deactivation manifest (CSV in vault) — list of N users with last-login + role
  4. PAUSE before deactivation — surface manifest in vault with handoff back to claude-code-or-aj for confirmation. Reclaim is reversible but visible to those users.

Acceptance: vault entry salesforce-p2.1-inactive-reclaim-manifest-2026-04-26.md with the deactivation candidate list + reversal procedure.

Task 3 — P2.2 Experience Cloud Guest Session Investigation (READ-ONLY)

  1. sf_list_experience_sites — enumerate active sites
  2. For each, query Network + NetworkMember + LoginEvent WHERE Application='Experience Cloud' AND CreatedDate=LAST_N_DAYS:7
  3. Identify which site is over-allocating guest sessions
  4. Recommend: scale-up purchase OR session timeout reduction (config change)

Acceptance: vault entry salesforce-p2.2-guest-session-investigation-2026-04-26.md naming the over-allocating site + remediation options.

What is NOT in scope this cycle

  • P1.2 storage purchase vs archive — AJ strategic decision (separate handoff)
  • P1.3 fullcopy refresh vs truncate — AJ strategic decision (irreversible refresh)
  • Tool bug patches (sf_monitor_governor_limits, sf_security_health_check) — file as MCP server v7.0.x bugs separately

Provenance

  • Triggered by: claude-code 2026-04-26 session-close audit
  • Reference: knowledge/critical/salesforce-mon-org-health-2026-04-26-3-p1-anomalies-storage.md
  • Mandate: AJ “no known thing pending” directive (2026-04-26)

Graph View