snapshot-report

Snapshot Report

Create the one report a CTO can read to understand what is going on and what must be communicated or prioritized.

Workflow

1. Resolve the project folder. Durable outputs live under <project>/knowledge/wiki/.
2. Prefer the deterministic helper:

dzcto snapshot "<project folder>"

Use --start YYYY-MM-DD --end YYYY-MM-DD for an explicit window, or --days N for a rolling window ending today.

3. Review the generated snapshot for:
   • TL;DR: overall status, the one thing to know, and the decision needed this week.
   • Communicate Up: CEO, founder, board, investor, or executive-facing points.
   • Communicate Down: team-facing expectations, decisions, focus, and process signals.
   • Decisions Needed: owner, deadline, and source for each decision.
   • Risks: deduped canonical risk-register rows only.
   • Top 5 Priorities: concrete next actions with done-when criteria.
   • Outcome Signals and Agent Activity Audit: product/readiness metrics and oversight coverage.
4. If the snapshot exposes stale or incorrect facts, update the canonical source: core/RISKS.md, core/DECISIONS.md, core/OPERATING_CADENCE.md, learning state, or the source report JSON. Do not edit generated HTML.
5. Refresh with dzcto refresh "<project folder>" after source changes.

Report Contract

The helper writes structured JSON and renders HTML under reports/snapshot/.

Expected JSON fields:

• executive_read
• window
• tldr
• changed_since_last_week
• metrics
• communicate_up
• communicate_down
• priorities
• application_state
• risks
• decisions
• operating_signals
• outcome_signals
• agent_activity_audit
• report_rollup
• sources

Standards

• This is a synthesis layer, not a new source of truth.
• Keep it short enough to read before a leadership meeting.
• Prefer sharp operating judgment over exhaustive coverage.
• Separate what should be communicated up from what should be communicated down.
• Make priorities actionable, with owners or next actions when known.
• Do not render a top-level follow-up-signals strip; those items are already covered in the Snapshot sections.
• Application State should use only the latest report for each report type, even when the snapshot window contains multiple runs.
• Report Rollup should use only the latest report for each report type; full history belongs in collapsed Sources.
• The main risk table should use canonical risk records only. Report risk rows are evidence/signals, not duplicate operating risks.
• Codebase Accountability must cover the snapshot window or the Agent Activity Audit should call out the rerun command.
• Keep Sources collapsed by default and clickable so provenance is available without crowding the report.