generate-plan-fact-report

generate-plan-fact-report

Language Detection

Determine the language of the user's request:

• If the request is in Russian → use templates with -ru suffix
• Otherwise → use templates with -en suffix

All output (headings, labels, comments, instructions) must match the detected language.

Triggers

Russian: «сформируй план-факт», «подготовь отчёт план/факт», «сравни план и факт» English: "generate plan-fact report", "plan vs actual", "variance report"

1. Input Data

Data	Required	Source	Notes
Project plan	yes	knowledge (project-plan.md)	Milestones, timelines, budget. If budget is missing — Section 3 is not generated
Actual data	yes	chat	User provides: actual milestone completion dates, actual costs by budget line, deliverable status. Free-form — agent structures it
Project charter	no	knowledge (project-charter.md)	For deliverables list (Section 1, scope). If unavailable — request the list from the user
Report type	no	chat	interim (default) or final. Affects header label only — logic is identical

If required data is missing — request it:

To generate the plan-vs-actual report, the following are needed:
1. Project plan (project-plan.md) — in knowledge or paste into chat.
2. Actual data — in any format:
   — milestone completion dates (completed, not completed, actual date);
   — actual costs by budget line;
   — which deliverables have been delivered and which have not.
Available: [list what is present].
Missing: [specify].

---

2. Execution Algorithm

1. Check input data. Verify project plan and actual data are present. Missing → request using the template in §1. Do NOT generate a placeholder report. Present → proceed to step 2.
2. Structure actual data. If free-form — organize across three dimensions: timelines (milestone dates), budget (costs by line), scope (deliverables). Note ambiguities and clarify with the user.
3. Check for data conflicts. If plan data in chat conflicts with project-plan.md — do NOT resolve independently. List the conflicts and ask which data is current. Exception: if the user explicitly corrects the plan in chat — apply the correction.
4. Determine report period. From actual data: project start → date of last recorded fact. If the user specified a period explicitly — use it.
5. Read the template plan-fact-report-ru.md or plan-fact-report-en.md from project knowledge (based on detected language).
6. Extract planned data from project-plan.md: milestones with dates, budget lines with amounts, deliverables list.
7. Compare plan vs. actual. For each row: calculate variance (absolute and %), assign status per §3 reference table.
8. Fill template sections per Section 3 rules.
9. Evaluate whether Section 4 is needed. Check statuses for all three dimensions independently: Timelines, Budget, Scope. Scope is a full dimension equal to timelines and budget. If at least one of the three has status attention or critical (including scope with incomplete deliverables) → fill Section 4. Only if all three are on track → remove Section 4 entirely.
10. Validate the result using the checklist in Section 5.
11. Present the result in chat:
    • What was done: "Plan-vs-actual report generated for period [dates] based on [list sources]."
    • Assumptions: list all (if any).
    • Key variances: brief (1–3 lines) — main discrepancies and their statuses.
    • Required action: "Approve, revise, or reject."
    • Do not include internal file paths. Filename only: "Report saved as plan-fact-report.md."
12. Wait for user response.
    • Approval → finalize the document.
    • Revisions → revise and present updated version. After the 3rd iteration — ask: "Continue revising or lock current version?"
13. After approval:
    • Offer the user text for log.md and project-state.md.
    • Always notify, regardless of report type and variance statuses: "Plan-vs-actual report approved. Next task available: generate-closure-report (project closure report). Start?"