start-sprint

name	start-sprint
description	Initialize a new sprint with GitHub issue parity, local tracker updates, and a testing/deployment strategy. Creates umbrella + sub-task issues, updates all local tracking docs, and defines validation requirements per sub-task and for the umbrella. Use at the START of a new sprint or workstream.
disable-model-invocation	true
allowed-tools	Bash(gh *) Bash(git *) Bash(grep *) Bash(ls *) Read Write Edit Glob Grep
argument-hint	<sprint-title> (e.g., "SkyPilot AWS Deployment")

You are assisting an ADEPT developer with initializing a new sprint (or multi-workstream feature). This skill ensures full parity between GitHub issues and local implementation tracking BEFORE any implementation begins.

Do not use emojis in any output. Enforce all rules in docs/development/AGENT_CODING_RULES.md throughout this skill execution. Do not include any AI agent references in issues or documents.

---

Phase 1: Gather Context

Read the active sprint status tracker:

ls -t docs/implementation-sprints/SPRINT_*_STATUS.md | head -1

Read the umbrella plan doc (if it exists) and identify:

• The overarching GitHub issue (#16 for FY26Q2)
• Existing workstreams and their issue numbers
• The sprint week and priority

Ask the developer:

1. Sprint/workstream title (use $ARGUMENTS if provided)
2. Number of sub-tasks/workstreams
3. Owner assignments (if known)
4. Priority and sprint week

---

Phase 2: Create GitHub Issues

2a. Umbrella Issue

Create an umbrella issue following the #25 or #33 pattern:

gh issue create \
  --title "<Sprint Title> -- <Subtitle>" \
  --body "$(cat <<'EOF'
## Overview
<1-2 sentence problem statement>

## Workstreams
- [ ] WS1: <title> (effort: X days)
- [ ] WS2: <title> (effort: X days)
- [ ] WSN: <title> (effort: X days)

## Dependency Graph

WS1 --> WS4 WS2 --> WS5 WS3 --> WS5

## Testing & Deployment Strategy
<How the umbrella will be validated end-to-end>

## Plan Documentation
- `docs/implementation-plans/<UMBRELLA_DOC>.md`
- `docs/implementation-plans/<subdir>/WS1_*.md`

## Sprint
- **Parent**: #16 (FY26Q2 Platform Optimizations)
- **Priority**: P1
- **Sprint**: W<N>+
EOF
)" \
  --label enhancement,infrastructure

Record the returned issue number.

2b. Per-Workstream Child Issues

For each workstream, create a child issue:

gh issue create \
  --title "WS<N>: <Workstream Title>" \
  --body "$(cat <<'EOF'
**Parent**: #<umbrella_number>
**Plan**: `docs/implementation-plans/<subdir>/WS<N>_*.md`
**Effort**: X days
**Dependencies**: <list or None>

## Scope
<Brief scope from plan doc>

## Testing Strategy
- **Offline validation**: <what can be validated without cloud resources>
- **Online validation**: <what requires credentials but no live resources>
- **Live deployment**: <what requires actual resource creation>
- **IaC parity**: <what infrastructure-as-code must be validated (CDK/Terraform/Pulumi)>

## Acceptance Criteria
- [ ] <criterion 1>
- [ ] <criterion 2>
EOF
)" \
  --label enhancement

Record each returned issue number.

---

Phase 3: Update Local Tracking

3a. Sprint Status Tracker

Add rows to the issue status table:

| #<umbrella> | <Sprint Title> | @<owner> | P1 | W<N>+ | Open (planning) | -- |
| #<ws1> | WS1: <title> | @<owner> | P1 | W<N> | In progress | -- |
| #<ws2> | WS2: <title> | -- | P1 | W<N>+ | Open | -- |

Update the issue hierarchy section and the summary count line.

3b. Umbrella Plan Doc

Add **GitHub Issue**: [#N](url) to the plan doc header. Add an issue-to-workstream mapping table in the Sprint Tracking section.

3c. Per-Workstream Plan Docs

Add to each WS doc header:

**GitHub Issue**: [#N](https://github.com/<org>/<repo>/issues/N)
**Parent**: [UMBRELLA_DOC.md](../UMBRELLA_DOC.md) ([#<umbrella>](url))

---

Phase 4: Define Testing & Deployment Strategy

CRITICAL: No sub-task is ready for commit without a validated testing strategy. The umbrella must define end-to-end deployment validation.

4a. Per-Workstream Testing Strategy

For each workstream, define and document in the plan doc:

Tier	What	How	Blocks Commit?
Offline	YAML/schema/lint validation	Containerized (no credentials)	YES
Online	Cloud connectivity, dry-run	With credentials, no resources	YES
Live	Actual deployment + health checks	Creates real resources	YES (first time)
IaC Parity	CDK/Terraform matches task YAML specs	Compare resource configs	YES

Each workstream MUST have:

• A Makefile validation target (e.g., make sky-validate-offline)
• A containerized test approach (never install tools directly on host)
• A CSP-specific validation path (AWS first, Azure/GCP later)
• Documented acceptance criteria that map to testable assertions

4b. Umbrella End-to-End Validation

The umbrella issue/doc must define:

1. Integration test: How workstreams interact (e.g., WS1 output feeds WS4)
2. CSP coverage: Which CSPs are validated (AWS primary, Azure/GCP planned)
3. IaC validation: CDK synth/diff, Terraform plan, Pulumi preview
4. Deployment pipeline: Full lifecycle test (provision -> deploy -> health check -> teardown)
5. Cost controls: Budget limits for live testing

4c. CSP Support Matrix

Document in umbrella plan doc:

CSP	IaC Tool	Status	Validation Target
AWS	CDK	Primary (first supported)	make sky-validate-aws
Azure	Pulumi	Planned	make sky-validate-azure (future)
GCP	Terraform	Planned	make sky-validate-gcp (future)

Note: Cloud-based servers use Ubuntu; on-premises VMs use Rocky Linux. Both OS families MUST be supported for parity with on-premises K8s clusters.

---

Phase 5: Update Tracking Docs

Update docs/CHANGELOG.md and docs/ROADMAP.md with the sprint initialization entry:

## Session <N>: <Sprint Title> Initialization (<Month Year>)
**Type**: Sprint planning
**GitHub Issues**: #<umbrella> (OPEN), #<ws1> (OPEN), ...

---

Phase 6: Verification Checklist

Before declaring sprint initialization complete, verify:

# All issues created
gh issue list --label enhancement --state open | grep "<sprint keyword>"

# Local tracker has all issue numbers
grep -c '#<umbrella>' docs/implementation-sprints/SPRINT_*_STATUS.md

# All plan docs have GitHub Issue headers
for ws in docs/implementation-plans/<subdir>/WS*.md; do
  grep -l "GitHub Issue" "$ws" || echo "MISSING: $ws"
done

# Testing strategy documented
grep -l "Testing Strategy\|Validation\|Acceptance Criteria" docs/implementation-plans/<subdir>/WS*.md

Present the checklist results. If any items fail, provide remediation steps.

---

Anti-Patterns (Avoid These)

• Starting implementation before GitHub issues exist
• Committing code without validation targets passing
• Creating a monolithic plan doc instead of per-workstream docs
• Skipping the testing strategy definition
• Not defining CSP parity requirements upfront
• Using host-installed tools instead of containerized validation
• Deploying to cloud without offline validation passing first
• Inspecting or mounting user home directories for credentials (~/.aws, ~/.ssh) — inject creds via env vars sourced from operational scripts (e.g., setup-claude-bedrock.sh)