sf-90-day-roadmap - SKILL.md Agent Skill

name: sf-90-day-roadmap description: Use when you need to turn a list of config debt findings into a phased, client-ready 90-day remediation plan with sprint-by-sprint actions. version: 1.0.0 author: SahirVhora license: MIT metadata: hermes: tags: [sapsf, roadmap, remediation, consulting, planning] related_skills: [sf-config-debt-scanner]

SF 90-Day Remediation Roadmap

Turn findings into action. Produce a prioritised 90-day plan the client can start executing on Monday.

When to Use

Client received findings and asks "what do we do now?"
Post-audit remediation planning
Building a project plan for configuration cleanup
Quarterly governance planning cycle
Proposal/SOW scope definition

Prerequisites

List of config debt findings with severities
Understanding of client's team capacity (how many people, how many hours)
Client's change management process (change freezes, release windows)

Workflow

Step 1: Score and Sort

Rate findings automatically:

mcp__sf_config_debt_scanner__sf_rate_findings(findings_json="...")

Sort by: severity → business impact → effort → dependency.

Step 2: Cluster into Sprints

Group related findings into thematic sprints:

Sprint theme: what all findings in this sprint share
Deliverables: 2-4 concrete outputs
Effort: person-days estimate
Dependencies: what must be fixed first
Owner: which team/role does the work

Step 3: Phase into 90 Days

Standard phasing:

Phase 1 (Days 1-30): Critical findings only. Quick wins that unblock later work.
Phase 2 (Days 31-60): High severity. More complex changes that depend on Phase 1.
Phase 3 (Days 61-90): Remaining medium/low. Documentation, governance setup, automation.

Step 4: Package for Client

Add:

Executive summary: one paragraph on what 90 days achieves
Risk register: what could go wrong and how to mitigate
Success metrics: how to know when each sprint is done
Governance cadence: weekly check-in, sprint review, phase gate

Edge Cases

100+ findings: triage ruthlessly. Top 20 make the 90-day plan. Rest go to a backlog.
Only low-severity findings: consider a 30-day plan instead; 90 days is overkill.
Mixed-source findings: findings from different scan tools need normalised scoring.
No effort data: use standard sizing (critical=2 days, high=1 day, medium=0.5 day, low=0.25 day) and note the assumption.
Change freeze during roadmap: Phase around the freeze window.
Findings that are by-design: some config debt is intentional. Exclude from remediation plan or flag as "accepted risk."

Common Pitfalls

Grouping unrelated findings into one sprint: Each sprint should have a clear theme the client can explain to their boss.
Ignoring dependencies: Fixing picklists before business rules will re-break the rules. Sequence correctly.
No success metric: "Fixed 12 findings" means nothing. "Payroll accuracy improved from 96% to 99.5%" means something.
Overpromising speed: 90 days is fast for config cleanup in large tenants. Be realistic about effort.
Not planning for regression testing: Every sprint needs UAT time. Build it into the plan.

Verification Checklist

Findings scored and sorted by severity and impact
Sprints are thematic with 2-4 deliverables each
Dependencies identified and sequenced
Success metrics defined per phase
Client capacity considered (not just ideal-world plan)
Change freeze windows accommodated
Governance cadence included