How long does the integration take?

Change one base URL in an OpenAI-compatible client, or install the Python or TypeScript SDK; the quickstart routes a first request in a few minutes.

Will this lock us into another vendor?

No. The router is Apache-2.0 open source, your provider keys stay yours, and the API is OpenAI-compatible, so leaving is a change of base URL.

What happens when a provider fails?

The router falls back to the next model your policy allows and records the fallback in the trace, so an outage costs a retry, not an incident.

plan:decompose - Prompt - OpenSmartRoute

Name: plan:decompose
Author: helalifaker

Prompt file imported from helalifaker/budfin (.codex/prompts/plan.decompose.md). Copyright stays with the author.

Internal command. Called by /workflow:run automatically.

Invoke the plan-decompose skill before any steps.

This command executes Phase 2 (DECOMPOSE) of the BudFin workflow in full.

Read .claude/workflow/STATUS.md first to confirm we are in Phase 2. If not, warn the user and stop.

Swarm Agent Requirement

Use TeamCreate to spin up a swarm before executing. Minimum 5 agents:

Agent	Role
Planner	Reads the plan-decompose skill catalog, determines creation order, drafts all 13 issue bodies before any `gh` calls
Implementer	Executes `gh issue create` and `gh project item-add` in dependency order (13→11→7→1→2→3→4→8→12→5→6→9→10)
Code Reviewer	After each batch of 3 epics, reviews created issue bodies on GitHub to confirm no placeholders, correct labels, correct dependency tables
QA	Runs the Step 6 quality gate across all 13 epics after creation; reports failures with the exact `gh issue edit` fix needed
Documentation	Updates STATUS.md Feature Progress table with real GitHub issue numbers after all epics pass QA

The Planner must complete its drafts before the Implementer begins. The Code Reviewer and QA run after implementation, in parallel with each other. The Documentation agent runs last.

Step 0: Pre-flight Check

Verify GitHub CLI access and repo connectivity before creating any resources:

gh auth status
gh repo view helalifaker/budfin --json name,url

If either command fails, stop and report the error to the user. Do not proceed until authentication is confirmed.

Step 1: Create Required Labels

gh label create epic --color "#7057ff" --description "Epic-level feature" 2>/dev/null || true
gh label create story --color "#0075ca" --description "Story-level task linked to an Epic" 2>/dev/null || true
gh label create "must-have" --color "#d93f0b" --description "MoSCoW: Must Have" 2>/dev/null || true
gh label create "should-have" --color "#e4e669" --description "MoSCoW: Should Have" 2>/dev/null || true
gh label create "could-have" --color "#0e8a16" --description "MoSCoW: Could Have" 2>/dev/null || true
gh label create blocker --color "#b60205" --description "Blocks other work" 2>/dev/null || true

Step 2: Create GitHub Projects Board

gh project create --owner helalifaker --title "BudFin Development" --format json

Save the project number from the output. Then confirm the Status field exists:

gh project field-list [PROJECT_NUMBER] --owner helalifaker

Note: GitHub Projects v2 creates a default Status field automatically with Backlog, In Progress, In Review, Done options.

Step 3: Create All 13 Epics

Use the epic catalog from the plan-decompose skill. Create epics in dependency-respecting order:

Order: 13 → 11 → 7 → 1 → 2 → 3 → 4 → 8 → 12 → 5 → 6 → 9 → 10

For each epic, use the full summary and dependency table from the skill. No placeholders — every field must be populated from the canonical catalog before creating the issue.

Issue creation pattern (repeat for each epic):

ISSUE_URL=$(gh issue create \
  --repo helalifaker/budfin \
  --title "Epic N: Feature Name" \
  --label "epic,must-have" \
  --body "## Summary

Epic N delivers [3-4 sentence summary from plan-decompose skill catalog. Must start with 'Epic N delivers...'. Must reference PRD section. Must state what EFIR staff can do after it ships.]

## MoSCoW Priority

Must Have

## Tier

MVP

## Acceptance Criteria Reference

Derived in Phase 4 via \`/plan:spec\` and stored in \`docs/specs/epic-N/\`.

## Stories

Will be created in Phase 4 via \`/plan:stories\`.

## Dependencies

| Epic | Feature | Reason |
|------|---------|--------|
| N    | Name    | Specific reason from skill catalog — not generic |

## Checklist

- [ ] Feature spec written (\`/plan:spec\`)
- [ ] Stories created (\`/plan:stories\`)
- [ ] All stories implemented and reviewed
- [ ] Epic closed")

ISSUE_NUMBER=$(echo "$ISSUE_URL" | grep -o '[0-9]*$')

gh project item-add [PROJECT_NUMBER] --owner helalifaker --url "$ISSUE_URL"

After creating each issue, record its number before moving to the next.

Step 4: Update STATUS.md

After all 13 epics are created, update .claude/workflow/STATUS.md:

Add a GitHub Issue column to the Feature Progress table
Populate each row with the actual issue number created
Format: | 1 | Enrollment & Capacity | — | Not Started | #[issue-number] |

Step 5: Report

Display a summary table:

Phase 2 — DECOMPOSE complete

| Epic | Feature                 | GitHub Issue |
|------|-------------------------|--------------|
| 1    | Enrollment & Capacity   | #N           |
| 2    | Revenue                 | #N           |
| 3    | Staffing (DHG)          | #N           |
| 4    | Staff Costs             | #N           |
| 5    | P&L Reporting           | #N           |
| 6    | Scenario Modeling       | #N           |
| 7    | Master Data Management  | #N           |
| 8    | Audit Trail             | #N           |
| 9    | Dashboard               | #N           |
| 10   | Version Management      | #N           |
| 11   | Authentication & RBAC   | #N           |
| 12   | Data Migration          | #N           |
| 13   | Infrastructure & CI/CD  | #N           |

Projects Board: https://github.com/orgs/helalifaker/projects/[N]

Next: /workflow:advance to gate-check Phase 2 and move to Phase 3 (SETUP).

Step 6: Quality Gate

After all 13 epics are created, verify each one passes these checks:

Summary does not contain [, ], "TBD", "placeholder", or "This feature"
Dependency table has at least one row (or explicitly states "None — foundational epic")
Issue has both epic label and a MoSCoW label (must-have or should-have)
Issue appears on the Projects board

If any epic fails a check, fix it with gh issue edit #N --body "..." before reporting complete.

Error Handling

If any gh command fails, display the error and continue with the next epic. Report all failures at the end so the user can fix them manually.

Install plan:decompose and it becomes one more candidate the router can pick - when it fits.

A listing is a routing target with a manifest: what it does, which domains it covers, what it costs and who publishes it. Once installed it sits beside your own models and tools, is scored like any other candidate for each request, and shows up in the trace when it wins. Ratings come from workspaces that installed it, one per account.

> **Internal command.** Called by `/workflow:run` automatically. Invoke the plan-decompose skill before any steps. This command executes Phase 2 (DECOMPOSE) of the BudFin workflow in full. Read `.claude/workflow/STATUS.md` first to confirm we are in Phase 2. If not, warn the user and stop. ## Swarm Agent Requirement Use `TeamCreate` to spin up a swarm before executing. Minimum 5 agents: | Agent | Role | | ------------- | ----------------------------------------------------------------------------------------------------------------------------------------- | | Planner | Reads the plan-decompose skill catalog, determines creation order, drafts all 13 issue bodies before any `gh` calls | | Implementer | Executes `gh issue create` and `gh project item-add` in dependency order (13→11→7→1→2→3→4→8→12→5→6→9→10) | | Code Reviewer | After each batch of 3 epics, reviews created issue bodies on GitHub to confirm no placeholders, correct labels, correct dependency tables | | QA | Runs the Step 6 quality gate across all 13 epics after creation; reports failures with the exact `gh issue edit` fix needed | | Documentation | Updates STATUS.md Feature Progress table with real GitHub issue numbers after all epics pass QA | The Planner must complete its drafts before the Implementer begins. The Code Reviewer and QA run after implementation, in parallel with each other. The Documentation agent runs last. ## Step 0: Pre-flight Check Verify GitHub CLI access and repo connectivity before creating any resources: ```bash gh auth status gh repo view helalifaker/budfin --json name,url ``` If either command fails, stop and report the error to the user. Do not proceed until authentication is confirmed. ## Step 1: Create Required Labels ```bash gh label create epic --color "#7057ff" --description "Epic-level feature" 2>/dev/null || true gh label create story --color "#0075ca" --description "Story-level task linked to an Epic" 2>/dev/null || true gh label create "must-have" --color "#d93f0b" --description "MoSCoW: Must Have" 2>/dev/null || true gh label create "should-have" --color "#e4e669" --description "MoSCoW: Should Have" 2>/dev/null || true gh label create "could-have" --color "#0e8a16" --description "MoSCoW: Could Have" 2>/dev/null || true gh label create blocker --color "#b60205" --description "Blocks other work" 2>/dev/null || true ``` ## Step 2: Create GitHub Projects Board ```bash gh project create --owner helalifaker --title "BudFin Development" --format json ``` Save the project number from the output. Then confirm the Status field exists: ```bash gh project field-list [PROJECT_NUMBER] --owner helalifaker ``` Note: GitHub Projects v2 creates a default Status field automatically with Backlog, In Progress, In Review, Done options. ## Step 3: Create All 13 Epics Use the epic catalog from the plan-decompose skill. Create epics in dependency-respecting order: **Order**: 13 → 11 → 7 → 1 → 2 → 3 → 4 → 8 → 12 → 5 → 6 → 9 → 10 For each epic, use the full summary and dependency table from the skill. No placeholders — every field must be populated from the canonical catalog before creating the issue. **Issue creation pattern** (repeat for each epic): ```bash ISSUE_URL=$(gh issue create \ --repo helalifaker/budfin \ --title "Epic N: Feature Name" \ --label "epic,must-have" \ --body "## Summary Epic N delivers [3-4 sentence summary from plan-decompose skill catalog. Must start with 'Epic N delivers...'. Must reference PRD section. Must state what EFIR staff can do after it ships.] ## MoSCoW Priority Must Have ## Tier MVP ## Acceptance Criteria Reference Derived in Phase 4 via \`/plan:spec\` and stored in \`docs/specs/epic-N/\`. ## Stories Will be created in Phase 4 via \`/plan:stories\`. ## Dependencies | Epic | Feature | Reason | |------|---------|--------| | N | Name | Specific reason from skill catalog — not generic | ## Checklist - [ ] Feature spec written (\`/plan:spec\`) - [ ] Stories created (\`/plan:stories\`) - [ ] All stories implemented and reviewed - [ ] Epic closed") ISSUE_NUMBER=$(echo "$ISSUE_URL" | grep -o '[0-9]*$') gh project item-add [PROJECT_NUMBER] --owner helalifaker --url "$ISSUE_URL" ``` After creating each issue, record its number before moving to the next. ## Step 4: Update STATUS.md After all 13 epics are created, update `.claude/workflow/STATUS.md`: - Add a GitHub Issue column to the Feature Progress table - Populate each row with the actual issue number created - Format: `| 1 | Enrollment & Capacity | — | Not Started | #[issue-number] |` ## Step 5: Report Display a summary table: ``` Phase 2 — DECOMPOSE complete | Epic | Feature | GitHub Issue | |------|-------------------------|--------------| | 1 | Enrollment & Capacity | #N | | 2 | Revenue | #N | | 3 | Staffing (DHG) | #N | | 4 | Staff Costs | #N | | 5 | P&L Reporting | #N | | 6 | Scenario Modeling | #N | | 7 | Master Data Management | #N | | 8 | Audit Trail | #N | | 9 | Dashboard | #N | | 10 | Version Management | #N | | 11 | Authentication & RBAC | #N | | 12 | Data Migration | #N | | 13 | Infrastructure & CI/CD | #N | Projects Board: https://github.com/orgs/helalifaker/projects/[N] Next: /workflow:advance to gate-check Phase 2 and move to Phase 3 (SETUP). ``` ## Step 6: Quality Gate After all 13 epics are created, verify each one passes these checks: 1. Summary does not contain `[`, `]`, "TBD", "placeholder", or "This feature" 2. Dependency table has at least one row (or explicitly states "None — foundational epic") 3. Issue has both `epic` label and a MoSCoW label (`must-have` or `should-have`) 4. Issue appears on the Projects board If any epic fails a check, fix it with `gh issue edit #N --body "..."` before reporting complete. ## Error Handling If any `gh` command fails, display the error and continue with the next epic. Report all failures at the end so the user can fix them manually.

{ "prompt": "> **Internal command.** Called by `/workflow:run` automatically.\n\nInvoke the plan-decompose skill before any steps.\n\nThis command executes Phase 2 (DECOMPOSE) of the BudFin workflow in full.\n\nRead `.claude/workflow/STATUS.md` first to confirm we are in Phase 2. If not, warn the user and stop.\n\n## Swarm Agent Requirement\n\nUse `TeamCreate` to spin up a swarm before executing. Minimum 5 agents:\n\n| Agent | Role |\n| ------------- | ----------------------------------------------------------------------------------------------------------------------------------------- |\n| Planner | Reads the plan-decompose skill catalog, determines creation order, drafts all 13 issue bodies before any `gh` calls |\n| Implementer | Executes `gh issue create` and `gh project item-add` in dependency order (13→11→7→1→2→3→4→8→12→5→6→9→10) |\n| Code Reviewer | After each batch of 3 epics, reviews created issue bodies on GitHub to confirm no placeholders, correct labels, correct dependency tables |\n| QA | Runs the Step 6 quality gate across all 13 epics after creation; reports failures with the exact `gh issue edit` fix needed |\n| Documentation | Updates STATUS.md Feature Progress table with real GitHub issue numbers after all epics pass QA |\n\nThe Planner must complete its drafts before the Implementer begins. The Code Reviewer and QA run after implementation, in parallel with each other. The Documentation agent runs last.\n\n## Step 0: Pre-flight Check\n\nVerify GitHub CLI access and repo connectivity before creating any resources:\n\n```bash\ngh auth status\ngh repo view helalifaker/budfin --json name,url\n```\n\nIf either command fails, stop and report the error to the user. Do not proceed until authentication is confirmed.\n\n## Step 1: Create Required Labels\n\n```bash\ngh label create epic --color \"#7057ff\" --description \"Epic-level feature\" 2>/dev/null || true\ngh label create story --color \"#0075ca\" --description \"Story-level task linked to an Epic\" 2>/dev/null || true\ngh label create \"must-have\" --color \"#d93f0b\" --description \"MoSCoW: Must Have\" 2>/dev/null || true\ngh label create \"should-have\" --color \"#e4e669\" --description \"MoSCoW: Should Have\" 2>/dev/null || true\ngh label create \"could-have\" --color \"#0e8a16\" --description \"MoSCoW: Could Have\" 2>/dev/null || true\ngh label create blocker --color \"#b60205\" --description \"Blocks other work\" 2>/dev/null || true\n```\n\n## Step 2: Create GitHub Projects Board\n\n```bash\ngh project create --owner helalifaker --title \"BudFin Development\" --format json\n```\n\nSave the project number from the output. Then confirm the Status field exists:\n\n```bash\ngh project field-list [PROJECT_NUMBER] --owner helalifaker\n```\n\nNote: GitHub Projects v2 creates a default Status field automatically with Backlog, In Progress, In Review, Done options.\n\n## Step 3: Create All 13 Epics\n\nUse the epic catalog from the plan-decompose skill. Create epics in dependency-respecting order:\n\n**Order**: 13 → 11 → 7 → 1 → 2 → 3 → 4 → 8 → 12 → 5 → 6 → 9 → 10\n\nFor each epic, use the full summary and dependency table from the skill. No placeholders — every field must be populated from the canonical catalog before creating the issue.\n\n**Issue creation pattern** (repeat for each epic):\n\n```bash\nISSUE_URL=$(gh issue create \\\n --repo helalifaker/budfin \\\n --title \"Epic N: Feature Name\" \\\n --label \"epic,must-have\" \\\n --body \"## Summary\n\nEpic N delivers [3-4 sentence summary from plan-decompose skill catalog. Must start with 'Epic N delivers...'. Must reference PRD section. Must state what EFIR staff can do after it ships.]\n\n## MoSCoW Priority\n\nMust Have\n\n## Tier\n\nMVP\n\n## Acceptance Criteria Reference\n\nDerived in Phase 4 via \\`/plan:spec\\` and stored in \\`docs/specs/epic-N/\\`.\n\n## Stories\n\nWill be created in Phase 4 via \\`/plan:stories\\`.\n\n## Dependencies\n\n| Epic | Feature | Reason |\n|------|---------|--------|\n| N | Name | Specific reason from skill catalog — not generic |\n\n## Checklist\n\n- [ ] Feature spec written (\\`/plan:spec\\`)\n- [ ] Stories created (\\`/plan:stories\\`)\n- [ ] All stories implemented and reviewed\n- [ ] Epic closed\")\n\nISSUE_NUMBER=$(echo \"$ISSUE_URL\" | grep -o '[0-9]*$')\n\ngh project item-add [PROJECT_NUMBER] --owner helalifaker --url \"$ISSUE_URL\"\n```\n\nAfter creating each issue, record its number before moving to the next.\n\n## Step 4: Update STATUS.md\n\nAfter all 13 epics are created, update `.claude/workflow/STATUS.md`:\n\n- Add a GitHub Issue column to the Feature Progress table\n- Populate each row with the actual issue number created\n- Format: `| 1 | Enrollment & Capacity | — | Not Started | #[issue-number] |`\n\n## Step 5: Report\n\nDisplay a summary table:\n\n```\nPhase 2 — DECOMPOSE complete\n\n| Epic | Feature | GitHub Issue |\n|------|-------------------------|--------------|\n| 1 | Enrollment & Capacity | #N |\n| 2 | Revenue | #N |\n| 3 | Staffing (DHG) | #N |\n| 4 | Staff Costs | #N |\n| 5 | P&L Reporting | #N |\n| 6 | Scenario Modeling | #N |\n| 7 | Master Data Management | #N |\n| 8 | Audit Trail | #N |\n| 9 | Dashboard | #N |\n| 10 | Version Management | #N |\n| 11 | Authentication & RBAC | #N |\n| 12 | Data Migration | #N |\n| 13 | Infrastructure & CI/CD | #N |\n\nProjects Board: https://github.com/orgs/helalifaker/projects/[N]\n\nNext: /workflow:advance to gate-check Phase 2 and move to Phase 3 (SETUP).\n```\n\n## Step 6: Quality Gate\n\nAfter all 13 epics are created, verify each one passes these checks:\n\n1. Summary does not contain `[`, `]`, \"TBD\", \"placeholder\", or \"This feature\"\n2. Dependency table has at least one row (or explicitly states \"None — foundational epic\")\n3. Issue has both `epic` label and a MoSCoW label (`must-have` or `should-have`)\n4. Issue appears on the Projects board\n\nIf any epic fails a check, fix it with `gh issue edit #N --body \"...\"` before reporting complete.\n\n## Error Handling\n\nIf any `gh` command fails, display the error and continue with the next epic.\nReport all failures at the end so the user can fix them manually.", "variables": [], "notes": "Phase 2 - Bulk-create all 13 BudFin GitHub Epics, required labels, and the GitHub Projects board with Backlog/In Progress/In Review/Done columns. Tools: Bash, Read, Edit.", "metadata": { "source": { "provider": "github-codex-prompts", "repository": "https://github.com/helalifaker/budfin", "path": ".codex/prompts/plan.decompose.md", "ref": "0ae459d12324d9c929baa7b09f783f5bc53911c6", "url": "https://github.com/helalifaker/budfin/blob/0ae459d12324d9c929baa7b09f783f5bc53911c6/.codex/prompts/plan.decompose.md", "key": "helalifaker/budfin/.codex/prompts/plan.decompose.md" } } }

plan:decompose

About

Swarm Agent Requirement

Step 0: Pre-flight Check

Step 1: Create Required Labels

Step 2: Create GitHub Projects Board

Step 3: Create All 13 Epics

Step 4: Update STATUS.md

Step 5: Report

Step 6: Quality Gate

Error Handling

Install plan:decompose and it becomes one more candidate the router can pick - when it fits.

What it costs to adopt - in engineering time.

One base URL

No vendor lock-in

Failover built in

Use it

Manifest

Reviews