Making the Process Work (Scrum/Kanban/SAFe)
Enterprise agile frameworks meet Solana reality: multisig deploys do not land at 3pm Friday, auditors return findings async, and Sev-1 interrupts sprints. Adapt ceremonies so process helps ship safe on-chain software.
Recipe
Quick-reference recipe card - copy-paste ready.
## Solana Definition of Done
- [ ] LiteSVM / anchor tests green
- [ ] IDL published if program changed
- [ ] Devnet deploy tx linked in ticket
- [ ] Runbook/monitor updated if ops impact
- [ ] Security checklist in PR descriptionWhen to reach for this:
- Adopting Scrum or Kanban on new Solana team
- SAFe PI planning includes blockchain workstream
- Retros find "done" stories without deploy
- Scaling past one squad
Working Example
## Kanban board columns
Backlog -> Spec/RFC -> In Dev -> In Review -> Devnet -> Ready for Mainnet -> Done
**WIP limit:** In Dev max 3 per engineer (program PRs are heavy)
**Expedite lane:** Sev-1 incident / audit Critical finding
- Preempts WIP; only IC can pull cardSprint planning adjustment:
- Reserve 20% capacity not pointed (incidents + reviews)
- Mainnet deploy stories split from devnet-complete stories
- Carry-over rule: if no devnet tx, story rolls automatically
What this demonstrates:
- Devnet deploy as explicit column, not imaginary "done"
- Expedite class for incidents
- Capacity reserve for Solana-specific interrupt work
Deep Dive
Scrum Adaptations
- Sprint goal: One measurable on-chain outcome (e.g., "deposit on devnet")
- Review demo: Explorer link + wallet flow, not slides only
- Retro: Include incident and near-miss review monthly
Kanban Adaptations
- Cycle time: Track spec -> devnet -> mainnet separately
- Blocked: "Waiting audit", "Waiting multisig" visible stickers
SAFe Adaptations
- Enabler stories: Kit migration, RPC failover - first-class on program board
- Architect sync: RFC pipeline reviewed in system sync
Gotchas
- Velocity as commitment - Story points ignore audit wait. Fix: Lead time metrics in SAFe/QBR.
- Done without devnet - "Merged" treated as shipped. Fix: DoD gate in Jira automation.
- No expedite policy - Every story "urgent". Fix: Only IC pulls expedite lane.
- Cross-team dependency invisible - Client blocked on IDL. Fix: Dependency tags on cards.
- Ceremony overload - No time to review
#[account]. Fix: Trim meetings; protect review time.
Alternatives
| Alternative | Use When | Don't Use When |
|---|---|---|
| Shape Up | Fixed appetite cycles | Heavy compliance traceability |
| Trunk-based | Small team, strong CI | Frequent breaking IDL |
| Waterfall audit gate | Regulated issuance | Fast iteration MVP |
FAQs
Scrum or Kanban for Solana?
Kanban if deploy cadence irregular; Scrum if 2-week rhythm works with product.
How point program stories?
Points reflect complexity + unknown; add deploy/audit subtasks for visibility.
PI planning length?
Include ecosystem enabler epics even if no user feature ships that PI.
Remote ceremonies?
Demo requires screen share of wallet + explorer - non-negotiable.
SAFe blockchain stream?
Dedicated ART for protocol; client squad consumes IDL releases.
Incidents in sprint?
Expedite; do not count toward velocity; track MTTR separately.
Retro formats?
Blameless; monthly incident/near-miss slot.
Tooling?
Linear, Jira, or GitHub Projects - enforce DoD fields.
Stakeholder in planning?
PM always; security monthly for harden stories.
Process change?
Experiment one sprint; measure lead time not satisfaction surveys only.
Related
- Delivery Best Practices - release DoD
- Estimation & Risk - sprint commitment input
- DORA Metrics for Web3 - measure process outcomes
Stack versions: This page was written for Agave 4.1.1, Solana CLI 3.0.10, Anchor 0.32.1, anchor-lang 0.32.1, Rust 1.91.1, @solana/kit 7.0.0, Surfpool 0.12.0, and LiteSVM 0.6.x.