Date: 2026-02-13
Authors: Dylan Fitzgerald + Claude (Opus 4.6)
Scope: Standalone tasks only (not subtasks). All numbers are total tasks including the current ~120.
The client currently has ~120 accepted standalone tasks on the Nebula platform and wants to reach 900 total.
Nebula is a simulated DevOps environment running a complete k3s Kubernetes cluster inside a single Docker container. It includes a full CI/CD pipeline (Gitea → Harbor → ArgoCD), an observability stack (Prometheus, Grafana, Loki, Jaeger), an Istio service mesh, and a microservices application ("Bleater," 8 Python/FastAPI services). The environment boots from a snapshot in ~60 seconds and runs ~60 pods across 15 namespaces in an air-gapped network. Tasks are standalone investigation/troubleshooting/implementation challenges that AI coding agents solve within this environment.
This analysis evaluates the viability of reaching four milestones — 200, 300, 500, and 900 total standalone tasks — across three scenarios:
- Nebula alone — expanding the existing platform
- Nebula + 1 additional environment — one new simulation platform
- Nebula + N additional environments — multiple new platforms
Each milestone is assessed for probability of achievement, estimated timeline, and critical dependencies.
We ran a systematic gap analysis across all four task categories (SRE, platform-engineering, devops, cloud-ops), scanning 185+ Discord threads across both client feedback channels, 37+ local task directories, and the full Nebula infrastructure manifest. Key findings:
- 43 raw gap ideas across categories, deduplicating to 28-30 unique viable ideas on the current platform
- 19 of those are low-risk (minimal overlap with existing tasks, clearly supported by Nebula infrastructure)
- 5 Nebula components have deployed infrastructure but zero task coverage (KEDA, GlitchTip, Statping-ng, Event Exporter, Maddy)
- 9+ SRE skill areas have no coverage at all (SLO/SLI management, postmortem processes, alert routing, graceful degradation, runbook automation, toil reduction, health probe design, LogQL alerting, GlitchTip error tracking)
- Platform-engineering is the densest category (~40 tasks), with remaining headroom mostly in platform operations (ArgoCD notifications, Harbor admin, KEDA triggers)
- Cloud-ops is the most heavily saturated (~55 tasks), with the biggest blind spots in workload controller lifecycle operations (StatefulSet, CronJob, DaemonSet, init containers)
These numbers define the carrying capacity of the current Nebula platform and inform all milestone estimates below.
pie title "Current Task Coverage by Saturation Level"
"Heavily Saturated (PostgreSQL, ArgoCD, CI/CD, Prometheus, Istio)" : 55
"Moderately Covered (Harbor, Redis, Helm, DNS, RabbitMQ)" : 40
"Low/No Coverage (KEDA, GlitchTip, Statping-ng, CronJobs, etc.)" : 25
The current 120 tasks are unevenly distributed. Five areas account for ~45% of all tasks (PostgreSQL operations, ArgoCD/GitOps, CI/CD pipelines, Prometheus observability, Istio service mesh), while 13 deployed components have zero task coverage.
| Expansion level | New tasks possible | Total ceiling | Engineering investment |
|---|---|---|---|
| None (current platform) | 35-50 | 155-170 | 0 |
| Light (utilize unused components) | 50-80 | 170-200 | 2-4 weeks |
| Moderate (add 3-5 components) | 80-130 | 200-250 | 2-3 months |
| Aggressive (add 7-10 components) | 150-230 | 270-350 | 4-8 months |
| Maximum (theoretical limit) | 200-330 | 320-450 | 8-14 months |
Why the ceiling exists: Each task runs in a fresh, isolated Nebula environment — there is no resource contention between tasks. The constraint is conceptual overlap. Tasks are rejected if solving one would automatically solve part of another, if two tasks produce the same deliverable, or if their graders check the same artifact. As task density on a given component increases, the remaining investigation space that doesn't trigger these overlap gates shrinks. A platform with ~35 components and a fixed application topology has a finite number of meaningfully distinct investigation patterns. The single-node architecture also eliminates multi-node scenarios (node affinity, topology spread, cluster federation) from the task space entirely.
Light expansion — no new components, just deeper task coverage:
- Bring 13 zero-coverage components (KEDA, GlitchTip, Statping-ng, Maddy, Event Exporter, CronJobs, Init Containers, etc.) to 2-4 tasks each
- Explore cross-component interaction tasks
- Add process/workflow tasks (postmortems, runbooks, toil audits)
Moderate expansion — add 3-5 new components:
| Component | Effort | Tasks enabled | Air-gap feasible? |
|---|---|---|---|
| Argo Workflows | 2-3 weeks | 10-15 | Yes |
| Cert-manager (advanced) | 1-2 weeks | 8-12 | Yes |
| Falco | 2-3 weeks | 8-12 | Yes |
| External Secrets Operator | 1-2 weeks | 5-10 | Yes |
| Velero (advanced) | 1-2 weeks | 5-8 | Yes (already present) |
Aggressive expansion — add 7-10 new components:
- Everything above, plus: Tekton Pipelines, Crossplane, Thanos/Mimir, Linkerd, Vault (advanced)
- Some candidates conflict with existing components (Linkerd vs. Istio, Tekton vs. Gitea Actions)
- Snapshot boot time and image size become constraints at 80+ pods, and some candidates conflict with existing components
A new simulation platform comparable to Nebula in scope — a self-contained, snapshot-bootable container with a full infrastructure stack, microservices application, and tooling. Building one is a major engineering project.
| Aspect | Nebula (reference) | New environment (estimate) |
|---|---|---|
| Engineering to build | 6+ months (already done) | 3-6 months |
| Component count | ~35 | 20-40 |
| Microservices | 8 (Bleater) | 5-10 (new app) |
| Task capacity | 320-450 | 200-400 |
| Boot time target | 60 seconds | 60-120 seconds |
mindmap
root((Additional<br>Environments))
Multi-Cluster Platform
2-3 k3s clusters
Mesh federation
Cross-cluster GitOps
Multi-cluster networking
Est. 200-300 tasks
Eng. 4-6 months
Cloud Infrastructure / IaC
Terraform + LocalStack
Simulated AWS services
Serverless patterns
VPC / security groups
Est. 200-350 tasks
Eng. 5-8 months
Legacy Modernization
VMs via QEMU/Firecracker
Ansible config mgmt
VM to container migration
Monolith decomposition
Est. 150-250 tasks
Eng. 5-8 months
Security Operations
Vault + Falco + SIEM
Compliance frameworks
Incident response toolchain
Forensics scenarios
Est. 100-200 tasks
Eng. 3-5 months
Key insight: Each environment targets a fundamentally different infrastructure stack, ensuring minimal cross-environment overlap. A multi-cluster platform tests federation skills that can't exist on single-node Nebula. A cloud/IaC platform tests Terraform skills that aren't relevant to K8s cluster operations.
| Environment | Conservative | Expected | Optimistic |
|---|---|---|---|
| Nebula (current, with aggressive expansion) | 270 | 320 | 450 |
| Multi-Cluster Platform | 200 | 250 | 300 |
| Cloud Infrastructure / IaC | 200 | 275 | 350 |
| Legacy Modernization | 150 | 200 | 250 |
| Security Operations | 100 | 150 | 200 |
flowchart LR
subgraph Nebula["Nebula Alone"]
N200["200 tasks<br>🟢 70-80%"]
end
subgraph Plus1["Nebula + 1 Environment"]
P200["200 tasks<br>🟢 95%+"]
end
subgraph PlusN["Nebula + N Environments"]
PN200["200 tasks<br>🟢 95%+<br>(overkill)"]
end
- Probability: 70-80%
- Required: ~80 new standalone tasks. Gap analysis found 28-30 directly; creative exploration adds 10-20; moderate expansion (2-3 new components) adds 25-35. Total pool: ~63-85.
- Timeline: 3-5 months FTE
- Risk factors: Some medium-overlap ideas may be rejected. Implementation attrition (~15%) reduces the pool.
- Critical dependency: Moderate Nebula expansion (at least 2 new components to reach 80 comfortably).
- Probability: 95%+
- Overkill for this target. The additional environment isn't needed; Nebula can likely reach 200 alone. Building a second environment just for this milestone would be a poor use of resources.
200 is achievable on Nebula alone with moderate expansion. This is the "high confidence" target. The gap analysis directly supports ~50 new tasks from identified gaps; the remaining ~30 come from creative exploration and 2-3 new components.
flowchart LR
subgraph Nebula["Nebula Alone"]
N300["300 tasks<br>🟡 25-35%"]
end
subgraph Plus1["Nebula + 1 Environment"]
P300["300 tasks<br>🟢 70-80%"]
end
subgraph PlusN["Nebula + N Environments"]
PN300["300 tasks<br>🟢 90%+<br>(comfortable)"]
end
- Probability: 25-35%
- Required: ~180 new standalone tasks. This pushes toward Nebula's ceiling even with aggressive expansion (7-10 new components = 150-230 new tasks). At this scale, later tasks are increasingly niche and overlap-adjacent.
- Timeline: 6-10 months FTE
- Risk factors: Overlap saturation — at 300 tasks on one platform, later ideas are increasingly niche and overlap-adjacent. Cross-task overlap checking becomes painful at this scale. Quality degrades on tail-end tasks as we exhaust the most natural investigation patterns.
- Critical dependency: Aggressive Nebula expansion AND high acceptance rate on niche tasks.
- Probability: 70-80%
- Split: Nebula contributes ~200 (with moderate expansion), new environment contributes ~100 (early-stage, low-hanging fruit).
- Timeline: 6-10 months FTE (3-5 months for Nebula tasks + 3-5 months overlapping for new environment build and early tasks)
- Critical dependency: Second environment engineering starts early (month 2-3).
- Probability: 90%+
- Comfortable margin. Each environment handles a smaller share.
300 is risky on Nebula alone but comfortable with one additional environment. The decision point: is the engineering investment in a second environment (3-6 months to build) worth the increased headroom? For 300 tasks, it depends on whether the client needs them quickly (Nebula-only is faster to start but harder to finish) or reliably (second environment provides margin).
flowchart LR
subgraph Nebula["Nebula Alone"]
N500["500 tasks<br>🔴 5-10%"]
end
subgraph Plus1["Nebula + 1 Environment"]
P500["500 tasks<br>🟡 40-50%"]
end
subgraph PlusN["Nebula + N Environments"]
PN500["500 tasks<br>🟢 70-80%<br>(2 additional)"]
end
- Probability: 5-10%
- Required: ~380 new standalone tasks. This exceeds Nebula's estimated ceiling of 320-450 total, meaning it requires hitting the absolute theoretical maximum with zero attrition. Practically impossible.
- Timeline: N/A (not achievable)
- Probability: 40-50%
- Split: Nebula ~250-300 (aggressive expansion), new environment ~200-250 (moderate maturity). Combined: 450-550.
- Timeline: 10-16 months FTE
- Risk factors: The second environment needs to reach moderate maturity (200+ tasks), which requires 6-8 months of active development after the initial 3-6 month build.
- Critical dependency: Second environment must be architecturally distinct enough to avoid cross-platform overlap. A multi-cluster or IaC platform provides the most differentiation.
- Probability: 70-80%
- Split: Nebula ~250, Env 2 ~150, Env 3 ~100. Each contributes from its strongest areas.
- Timeline: 12-18 months (environments can be built in parallel by different teams)
- Critical dependency: Team scaling — 2 people can't build 2 new environments while also producing tasks. Needs 4-6 people.
500 is not achievable on Nebula alone. It requires at least one additional environment, and two makes it comfortable. This is where the project transitions from "task authoring" to "platform engineering + task authoring." The bottleneck shifts from idea generation to environment construction.
flowchart LR
subgraph Nebula["Nebula Alone"]
N900["900 tasks<br>🔴 0%<br>(impossible)"]
end
subgraph Plus1["Nebula + 1 Environment"]
P900["900 tasks<br>🔴 <5%"]
end
subgraph Plus2["Nebula + 2 Environments"]
P2_900["900 tasks<br>🟡 20-30%"]
end
subgraph Plus3["Nebula + 3 Environments"]
P3_900["900 tasks<br>🟡 40-55%"]
end
subgraph Plus4["Nebula + 4+ Environments"]
P4_900["900 tasks<br>🟢 60-70%"]
end
- Probability: ~0%
- The platform's absolute ceiling is 320-450 standalone tasks. 900 is mathematically impossible on Nebula alone.
- Probability: <5%
- Combined ceiling: ~450-700. Even at the optimistic end, this falls short of 900. Both platforms would need to hit their absolute theoretical maximums.
- Probability: 20-30%
- Split: Nebula ~300-350, Env 2 ~250-300, Env 3 ~200-250. Combined: 750-900.
- Timeline: 18-24 months
- Risk factors: This is at the combined ceiling. All three platforms need aggressive expansion and near-maximum utilization. Any one platform underperforming kills the target.
- Team size: 5-8 people (2 per environment + 1-2 on cross-cutting coordination)
- Probability: 40-55%
- Split: Nebula ~300, Env 2 ~250, Env 3 ~200, Env 4 ~150. Combined: 900.
- Timeline: 18-24 months (environments built in parallel)
- Risk factors: Coordination overhead across 4 platforms. Cross-platform overlap checking. Consistent quality standards.
- Team size: 6-10 people
- Probability: 60-70%
- Provides comfortable margin. Each platform contributes 150-250 tasks without pushing any to its ceiling.
- Timeline: 20-30 months
- Team size: 8-12 people
900 standalone tasks is a multi-platform, multi-team, multi-year program. It requires 3-4 simulation platforms with distinct infrastructure stacks, a team of 6-10+ task authors and platform engineers, and 18-24 months of sustained effort. It is not achievable by two people on one platform in any timeframe.
xychart-beta
title "Probability of Reaching Milestone by Scenario"
x-axis ["200", "300", "500", "900"]
y-axis "Probability (%)" 0 --> 100
line "Nebula alone" [75, 30, 7, 0]
line "Nebula + 1 env" [95, 75, 45, 4]
line "Nebula + 2+ env" [95, 90, 75, 45]
Legend: Top line = Nebula + 2+ env | Middle line = Nebula + 1 env | Bottom line = Nebula alone
| Milestone | Nebula Alone | + 1 Environment | + 2 Environments | + 3+ Environments |
|---|---|---|---|---|
| 200 | 🟢 70-80% | 🟢 95%+ | 🟢 95%+ | 🟢 95%+ |
| 300 | 🟡 25-35% | 🟢 70-80% | 🟢 90%+ | 🟢 95%+ |
| 500 | 🔴 5-10% | 🟡 40-50% | 🟢 70-80% | 🟢 85%+ |
| 900 | 🔴 ~0% | 🔴 <5% | 🟡 20-30% | 🟡 40-55% |
| Milestone | Nebula Alone | + 1 Env | + 2 Env | + 3+ Env |
|---|---|---|---|---|
| 200 | 3-5 mo | — | — | — |
| 300 | 6-10 mo | 6-10 mo | — | — |
| 500 | N/A | 10-16 mo | 12-18 mo | — |
| 900 | N/A | N/A | 18-24 mo | 20-30 mo |
| Milestone | Minimum team | Recommended team |
|---|---|---|
| 200 | 2 (us) | 2-3 |
| 300 | 2-3 | 3-5 |
| 500 | 4-6 | 5-8 |
| 900 | 6-10 | 8-12 |
-
For 200: Proceed on Nebula alone. We have the ideas, tooling, and capacity. Start immediately.
-
For 300: Begin Nebula work immediately while scoping a second environment. The environment decision should be made within 4-6 weeks based on Nebula progress.
-
For 500: Commit to a second environment early. The IaC/Cloud platform (Terraform + LocalStack) or Multi-Cluster platform offer the most differentiated task space. Start environment engineering in month 2.
-
For 900: This requires a program-level commitment — multiple platforms, a scaled team, and 18-24 months. We should present this honestly to the client with the data above and discuss whether 500-600 high-quality tasks (achievable with 2 environments) better serves their needs than 900 tasks of varying quality.