olepfm-session-manager

Imported from dime-worldbank/outcome-led-pfm-toolkit (skills/olepfm-session-manager/SKILL.md). Install upstream with npx skills add dime-worldbank/outcome-led-pfm-toolkit --skill olepfm-session-manager. Copyright stays with the author.

OLePFM Sector Reform Design: Session Manager

This document is the orchestration guide for producing a complete OLePFM Sector Reform Design Report in a single AI session. Upload it together with the ten prompt files and the reference documents at the start of the session.

TWO WAYS TO USE THIS SKILL

Workshop mode is an overlay on the report workflow, not a separate process: the instructions to the AI below (sequential execution, one prompt at a time, pause at every ✏️ point, never invent placeholder values) apply in both modes.

INSTRUCTIONS TO THE AI

You are assisting with the production of an OLePFM Sector Reform Design Report. Follow these rules throughout the session:

1. Sequential execution only. Work through the five stages — Sources, Step 1, Step 2, Step 3, Compilation and QA — in strict order. Do not begin a new stage until the user explicitly says "proceed to Step [n]" (or "proceed to Compilation") or equivalent.

2. One prompt at a time. Run each numbered prompt individually; after each output, wait for the user before the next. Do not chain prompts.

3. Pause at every ✏️ intervention point. After every prompt that has a ✏️ marker, output a block in this shape before stopping — plain Markdown, not a > blockquote (which breaks the checkboxes):

   ✏️ Review the output — focus on these points

   [topic] — [one line on why it matters, drawn from the instruction text before the prompt].

   • [check 1]
   • [check 2]
   • [check 3]

   Say "proceed" when satisfied, or provide corrections.

   Show all checks as one checklist, from the skill's ✏️ block.

4. Declare stage completion. At the end of each stage, output:

   STEP [n] COMPLETE (or SOURCES COMPLETE, COMPILATION AND QA COMPLETE) Please review the completion gate below before I proceed to [next stage].

5. Record inter-step data. At the specified points, ask the user to confirm the recorded values before proceeding. These values carry forward across stages.

6. Do not invent or skip. If a prompt asks you to fill in a placeholder such as [number] or [paste bottleneck title], pause and ask the user for the value rather than inventing one.

SESSION SETUP

Complete the two setup parts in manager-setup.md (in references/) before starting the Sources stage. Setup 1 verifies that the Report Template and the Synthesis Handbook are accessible. Setup 2 establishes and locks the session context (country, sector, development outcome and public sector result), using the Sector Outcome Reference in that file; it also says when a sector uses variant prompts. Do not begin Sources until the context is confirmed.

Skill Files

The skill files (00-source-prep.md through 04c-cross-document-qa.md) are in the knowledge base's skills folder, accessible by default — no upload required. In this repository they are the files in the references/ folder beside this SKILL.md.

Prompt numbering. Prompts are numbered by stage and run in order: S-1 to S-4 (Sources), 1-01 to 1-15 (Step 1), 2-01 to 2-10 (Step 2), 3-01 to 3-14 (Step 3), C-01 to C-13 (Compilation) and QA-1 to QA-5 (final cross-document checks). Optional quality checks at the end of a step are numbered 1-Q1, 2-Q1 and so on. Inside each file, prompts are grouped under the OLePFM sub-step they serve (for example Sub-step 1.4, Map the sector policy, institutional and public finance context). A sector variant of a prompt keeps the standard number with a prefix: ER-1-05 is the economic-resilience variant of Prompt 1-05.

SOURCES: Source Preparation

Skill file: 00-source-prep.md

Run Prompts S-1 through S-4 from 00-source-prep.md in sequence.

Sources Completion Gate

Before advancing to Step 1, confirm all of the following:

• The most important recent analytical reports for [country] and [sector] are included
• Official government budget documents and sector strategy documents are identified
• International data sources (WHO, World Bank, IMF) are included
• Budget and expenditure data — including subnational data — are identified
• Political economy and reform history sources are identified (relevant to Step 3)
• Capacity and digital systems sources are identified (relevant to Step 3)
• Every source sits in a single running table, numbered sequentially and unbroken across S-1 to S-4
• Any significant source gaps are flagged before proceeding

SOURCES COMPLETE — confirm sources before proceeding to Step 1.

STEP 1: Chapters 1 & 2 and Annex Step 1

Skill files: 01a-chapters1-2.md, then 01b-annex1.md

Run all prompts in 01a-chapters1-2.md in sequence (Prompts 1-01 through 1-11), then continue with 01b-annex1.md (Prompts 1-12 to 1-15).

Economic resilience or another cross-cutting fiscal outcome (macro-fiscal stability, fiscal or debt sustainability): run Prompts 1-01 to 1-03 as normal, then replace Prompts 1-04 to 1-10 with ER-1-04 to ER-1-10 from 01a-variant-economic-resilience.md (ER-1-04 to ER-1-06) and 01a-variant-economic-resilience-2.md (ER-1-07 to ER-1-10), then return to Prompt 1-11 and 01b-annex1.md. Also run the extra check ER-1-Q5. Record the variant's three template deviations (fiscal-outcome swim lanes; fiscal-cycle narrative and diagram instead of the financial-flows table and diagram; macro-fiscal charts instead of sector-spending charts) for the consistency audit (Prompt C-01).

Step 1 Completion Gate

Before advancing to Step 2, confirm all of the following:

Chapter 1

• Outcome is framed at the right level of specificity for the country context
• Purpose statement correctly describes the process and intended use of the report
• Section structure roadmap matches the structure to be produced

Section 2.1 — Outcomes and Public Sector Results (KEY)

• Headline development outcome indicators are accurate and sourced from authoritative data
• Public sector results reflect actual delivery system performance in [country]
• Framing is consistent with government's stated policy objectives

Section 2.2 — Public Sector Challenges (MOST CRITICAL in Step 1)

• Challenges are the most structurally important constraints — not symptoms
• Each challenge is distinct and covers a different dimension of the delivery problem
• Challenges progress from the point of delivery upward through the system
• No challenge mentions public finance or PFM
• Number of challenges (3–5) is appropriate for the complexity of the sector
• Final challenge titles confirmed

Section 2.3 — Context

• All significant organizations are included at correct levels; diagram reviewed
• All significant financing channels are identified; data is accurate; diagram reviewed (economic-resilience variant: fiscal-cycle narrative and diagram reviewed instead)
• PFM systems description covers all six dimensions (budget formulation, wage execution, non-wage execution, intergovernmental transfers, procurement, audit/accountability, digital systems) (economic-resilience variant: the four macro-fiscal capabilities instead, and the three template deviations recorded)
• Feasibility assessment addresses financial affordability, institutional capability, and stakeholder commitment with specific evidence

Annex Tables 1.1 to 1.3

• Table 1.1 consistent with Sections 2.1 and 2.2
• Table 1.2 covers all organizations; no significant actor omitted
• Table 1.3 lists the same financing channels, values and problems as the Section 2.3.2 flows table

Inter-Step Record — Step 1 → Step 2

Record the confirmed challenge titles before proceeding:

Challenge 1: 
Challenge 2: 
Challenge 3: 
Challenge 4: 
Challenge 5 (if applicable): 

STEP 1 COMPLETE — confirm challenges recorded before proceeding to Step 2.

STEP 2: Chapter 3 and Annex Step 2

Skill file: 02-chapter3-annex2.md

Run all prompts in 02-chapter3-annex2.md in sequence.

Note: The session context and source list are already in context — begin directly with Prompt 2-01.

Step 2 Completion Gate

Before advancing to Step 3, confirm all of the following:

Section 3.1 — Role of Public Finance

• Four-role assessment is specific to [country]'s delivery model — not generic
• Actual role descriptions are analytically honest and supported by evidence
• Section correctly foreshadows the bottlenecks that follow

Bottleneck Analysis — Annex Table 2.2 (KEY)

• Five-why analysis completed for each public sector challenge
• Bottlenecks are root causes, not symptoms or restatements of the challenge
• Type classifications (Technical / Institutional / PFM) are correct
• Impact (H/M/L) and feasibility (H/M/L) ratings are consistent across all bottlenecks

Consolidated Bottlenecks and Prioritization (KEY)

• Thematic groupings are appropriate for the country and sector context
• No bottleneck is so broad as to be unactionable
• Priority ranking reflects real political economy — not just technical assessments
• Any out-of-scope bottlenecks flagged as parallel tracks

Sections 3.2.1 and 3.2.2

• Bottleneck codes consistently cross-referenced (B1.1, B2.3, etc.)
• Each priority bottleneck has a concrete, observable change objective
• Annex Table 2.3 consistent with Section 3.2.2

Inter-Step Record — Step 2 → Step 3

Record the confirmed priority bottlenecks before proceeding:

Number of priority bottlenecks: 
Bottleneck 1: [title] | Change objective: 
Bottleneck 2: [title] | Change objective: 
Bottleneck 3: [title] | Change objective: 
Bottleneck 4: [title] | Change objective: 
Bottleneck 5: [title] | Change objective: 
Bottleneck 6: [title] | Change objective: 

STEP 2 COMPLETE — confirm bottlenecks recorded before proceeding to Step 3.

STEP 3: Chapters 4 & 5 and Annex Step 3

Skill files: 03a-reform-results.md, then 03b-stakeholders-systems-steps.md, then 03c-conclusion-compilation.md

Run all prompts in 03a-reform-results.md in sequence (Prompts 3-01, 3-02 × n, 3-03, 3-04 × n, 3-05), then 03b-stakeholders-systems-steps.md (Prompts 3-06 to 3-10, 3-11 × n, 3-12), then 03c-conclusion-compilation.md (Prompts 3-13 and 3-14).

Note: The session context and source list are already in context — begin directly with Prompt 3-01.

Critical sequencing: Complete and confirm Annex Table 3.1 (Prompts 3-02 × n and 3-03, then check 3-Q5) before beginning Section 4.1 drafts (Prompts 3-04 × n and 3-05). The reform results in Table 3.1 are the authoritative source for all Section 4.1 content.

Step 3 Completion Gate

Before advancing to Compilation and QA, confirm all of the following:

Annex Table 3.1 (must be confirmed before Section 4.1)

• Every sub-bottleneck has a row with specific, evidence-based underlying causes, distilled from a multi-branch five-whys into root-cause families
• Every root-cause family is addressed by a reform and every reform traces to a cause (check 3-Q5 passed)
• All key stakeholders named at organizational level — not generic actor types
• Every reform result is a concrete, measurable end-state with a feasibility rating
• Table stands alone as a self-explanatory reference document

Section 4.1 — Reforms to Deliver Change Objectives (KEY)

• Total number of reform results is within range (ideally 15–25)
• Feasibility ratings consistent with Annex Tables 2.2 and 2.3
• All cross-references to Annex Table 3.1 are correctly placed
• Change objective sub-section titles use the positive outcome statement — not the bottleneck problem description

Section 4.2 — Stakeholder Strategy

• Main political economy challenges identified and addressed with specific strategies
• No major reform result left without an identified authorizer and team leader
• Annex Tables 3.2 and 3.3 correctly cross-referenced

Annex Table 3.5 — Key Steps

• Exactly three steps per reform result, correctly sequenced
• Timing consistent with feasibility ratings
• No single organization has an unrealistic number of simultaneous lead responsibilities
• Implementation horizon applied consistently throughout

Chapter 5 — Conclusion

• Feasibility counts (H/M/L) accurate
• Low-feasibility reforms named explicitly with specific political economy strategies
• Scope boundary correctly and completely stated
• Concluding paragraph connects reform objectives back to the development outcome

Inter-Step Record — Step 3 → Compilation and QA

Record the following before proceeding:

Total reform results: 
High-feasibility results: [list result numbers]
Medium-feasibility results: [list result numbers]
Low-feasibility results: [list result numbers]
Implementation horizon: [start year] to [end year]
Authors: 
Report date: 

STEP 3 COMPLETE — confirm values recorded before proceeding to Compilation and QA.

COMPILATION AND QA: Final Compilation and Quality Assurance

Skill files: 04a-audit-and-part1.md, then 04b-part2-and-annex.md, then 04c-cross-document-qa.md

Start with the consistency audit (Prompt C-01) before any compilation step. Resolve all identified discrepancies before proceeding to Prompt C-02.

Economic-resilience variant: if the Section 2.3 variant prompts were used in Step 1, give the audit the three recorded template deviations (fiscal-outcome swim lanes; fiscal-cycle narrative and diagram in place of the financial-flows table and diagram; macro-fiscal charts in place of sector-spending charts) so it treats them as intended substitutions, not inconsistencies.

Run all prompts in 04a-audit-and-part1.md in sequence (Prompts C-01 to C-05), then 04b-part2-and-annex.md (Prompts C-06 to C-13), then 04c-cross-document-qa.md (Prompts QA-1 to QA-5).

Compilation and QA Completion Gate

Before treating the report as complete, confirm all of the following:

Pre-Compilation Audit

• All reform result codes, titles, and feasibility ratings are identical wherever they appear
• All bottleneck codes (e.g. B2.3) match between Chapter 3 and Annex Table 2.2
• Public sector challenge titles are consistent across all chapters and annexes
• Stakeholder names are consistent throughout (abbreviations follow a single convention)
• All citations are hyperlinked to their source, e.g. (World Bank, 2010)

Main Report Part 1

• Executive summary accurately represents the entire report in plain language
• Table of contents is complete and correctly numbered
• All cross-references to Part 2 and Annex are correct

Main Report Part 2

• Four-role structure (3.1.1–3.1.4) maintained with standard headings
• All reform result numbers in Chapter 4 match Annex Tables 3.1 and 3.5
• Chapter 5 feasibility summary matches Section 4.1 summary table
• Bibliography complete; every in-text citation has a bibliography entry

Annex

• Tables 1.1 to 1.3 consistent with Chapter 2
• Tables 2.2 and 2.3 consistent with Chapter 3; every bottleneck in 2.2 accounted for in 2.3
• Table 3.1 consistent with Section 4.1
• Table 3.5 consistent with Sections 4.2 and 4.3; three steps per reform result throughout

Cross-Document QA (Prompts QA-1 to QA-5)

• Reform result codes consistent across all three documents
• Challenge titles consistent across all three documents
• Outcome and spending data internally consistent
• Annex completeness confirmed
• Executive summary and Chapter 5 plain language reviewed

REPORT COMPLETE when all boxes are checked.

Context Length Advisory

If the AI loses track (wrong codes, inconsistent challenge titles), ask it to list the Section 2.2 challenges verbatim. If drift is severe, save the Step 1–3 outputs, restart the session, re-upload, and jump to Compilation and QA.