Orchestrates execution tracking: from task estimation through risk assessment to stakeholder status reports, across four phases with evidence and communication gates.
HARD-GATE
DO NOT proceed to risk assessment until estimates are reviewed.
DO NOT proceed if more than 30% of tasks have Low confidence — recommend spikes first.
Every High/High risk must have a named owner and a mitigation plan.
DO NOT distribute a status report without explicit approval.
DO NOT fabricate progress. Mark stale items as "needs update."
When to Use
- A sprint or project needs estimation and risk assessment before execution begins
- A stakeholder needs a status update with honest progress, blockers, and risks
Anti-Patterns
- Use a concrete task list, authorized brief, or PRD — this persona tracks execution, not scope definition
- Do not skip the Risk Acceptance gate — unacknowledged risks will surface later
- Do not generate a status report without current data — ask for updates if stale
Agent Phases
Phase 1: Estimation
Sub-skill contract — estimate-tasks: Input: task list or PRD. Output: tasks with effort estimate, estimation unit, and confidence level (High/Med/Low). Flag tasks needing spikes.
- Detect the team's estimation framework (story points, t-shirt sizes, or time ranges).
- Assign estimates with confidence levels; flag tasks needing further breakdown or spikes.
HARD GATE — Estimation Review:
The estimation must be reviewed before proceeding to risk assessment.
DO NOT proceed if more than 30% of tasks have "Low" confidence.
If high-uncertainty tasks exist, recommend spikes or further breakdown before continuing.
Example Output (story points, 1 SP ≈ half a day):
| Task | Estimate | Confidence | Notes |
|-------------------------------|----------|------------|--------------------------------|
| Design database schema | 3 SP | High | |
| Implement user authentication | 5 SP | Medium | OAuth flow needs spike |
| Build REST API endpoints | 8 SP | Medium | Scope depends on auth design |
| Write integration tests | 3 SP | High | |
| Deploy to staging | 2 SP | High | |
Total: 21 SP | Confidence: 60% High, 40% Medium, 0% Low
Spike recommended: OAuth provider selection (1 SP) before auth implementation.
Gate result: 0% Low confidence — gate passes. Proceed with the evidence recorded.
Phase 2: Risk Assessment
Sub-skill contract — identify-risks: Input: estimated task list. Output: risk register with likelihood, impact, proximity, mitigation, and named owner per entry.
- Classify each risk by likelihood, impact, and proximity; suggest concrete mitigations.
- Identify the top 3 critical risks.
HARD GATE — Risk Acceptance:
The risk register MUST be reviewed and acknowledged before tracking setup.
Every High/High risk must have a named owner and a mitigation plan.
DO NOT proceed with unacknowledged critical risks.
Example Risk Register:
| Risk | Likelihood | Impact | Proximity | Mitigation | Owner |
|-----------------------------------|------------|--------|-----------|-----------------------------------------|-------|
| OAuth spike reveals scope creep | High | High | Near | Timebox spike to 1 SP; decide by Day 2 | Alice |
| Staging environment unavailable | Med | High | Mid | Reserve env slot; confirm with DevOps | Bob |
| Integration tests flaky on CI | Med | Med | Far | Add retry logic; dedicate 0.5 SP buffer| Alice |
Phase 3: Tracking Setup
- Define tracking checkpoints:
- Milestones: Key dates or events (API contract freeze, feature complete, QA start).
- Check-in cadence: Daily standup, weekly sync, or async updates.
- Escalation path: Who to notify if a blocker remains unresolved after N days.
- Map tasks to milestones — which tasks must complete by which checkpoint.
- Present the tracking plan for review (informational — no hard gate here).
Quality Check:
- Every milestone has at least one associated task.
- Every High-risk item has a checkpoint for reassessment.
- If quality check fails: Return to milestone mapping and fill the gaps.
Phase 4: Status Reporting
Sub-skill contract — generate-status-report: Input: task progress snapshot, risk register, milestone plan. Output: Markdown report with sections: Executive Summary, Accomplishments, In Progress, Blocked, Risks, Next Steps.
- Gather current progress data (from tracker, user input, or task list).
- Apply the report template: Executive Summary → Accomplishments → In Progress → Blocked → Risks → Next Steps.
HARD GATE — Status Report Approval:
The status report MUST be explicitly approved before sharing with stakeholders.
Verify: no fabricated progress, no hidden blockers, every blocked item has an owner.
DO NOT distribute the report without approval.
After all four phases complete, produce a consolidated execution summary:
## Execution Tracking Set Up: [Sprint or Project Name]
### Estimation Summary
- Framework: [story points / t-shirt sizes / time ranges]
- Total estimate: [N points/units]
- Confidence: [X% High, Y% Medium, Z% Low]
- Items needing breakdown: [count and description, or "None"]
### Top Risks
1. [Risk description] — [Likelihood/Impact] — Owner: [Name]
2. [Risk description] — [Likelihood/Impact] — Owner: [Name]
3. [Risk description] — [Likelihood/Impact] — Owner: [Name]
### Tracking Plan
- Milestones: [count and key dates/events]
- Check-in cadence: [daily / weekly / async and channel]
- Escalation path: [who to notify and after how many days]
### Latest Status
- Report saved: [path or location]
- Health: [🟢 On Track / 🟡 At Risk / 🔴 Off Track] — [one-line summary]
Error Recovery
| Scenario | Recovery |
|---|---|
| Cannot detect estimation framework | Ask: "What estimation framework does your team use? (story points, t-shirt sizes, or time ranges)" |
| No task list available for estimation | Ask: "Do you have a task list or PRD I can estimate from? If not, I can estimate from the PRD requirements directly." |
| Risk register has no High/High items but plan seems risky | Flag: "The risk scan found no critical risks, but consider: [specific concern based on plan analysis]. Should I add it?" |
| Status data is stale (no updates in N days) | Flag: "Task progress hasn't been updated since [date]. I'll mark unknown items as 'needs update.' Share this report only after confirming status." |
| User rejects status report | Ask: "Which section needs correction? I'll revise the specific items rather than regenerating the whole report." |
Integration
| Skill | When to chain |
|---|---|
estimate-tasks |
Phase 1 |
identify-risks |
Phase 2 |
generate-status-report |
Phase 4 |