Decompose
Create well-defined JIRA artifacts (Features or Initiatives) from a strategy-level source issue (RFE, Outcome, OCPSTRAT, or any issue with linked RFEs). Triggers on: /rfe:decompose, "decompose RFE", "create features from RFE", "decompose outcome into features", "create features from OCPSTRAT", "create initiative from RFE"From its SKILL.md
npx -y skills add jhjaggars/rfe --skill decomposeAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
8.5 KB, as published. Nobody here has run it
decompose
You take a strategy-level JIRA issue — an RFE, Outcome, OCPSTRAT issue, or similar — gather deep context from it and all its linked issues, ask targeted interview questions to fill gaps, then create well-defined Feature or Initiative issues in the appropriate JIRA project.
Read these reference files now before proceeding:
references/feature-definition.md— Feature template, project routing, custom field IDs, andaclicreation patternreferences/artifact-hierarchy.md— when to create a Feature vs. Initiative vs. Outcome, lifecycle statuses, and project routingreferences/artifact-templates.md— Initiative and Outcome body templates andacliJSON payloadsreferences/wiki-markup.md— Jira wiki markup reference (for understanding existing issue descriptions; create new issues using ADF)
Phase 1: Gather Context
Step 1: Fetch the source issue.
Run `acli jira workitem view <KEY> --json` (or appropriate acli command) and parse the JSON output directly.
Step 2: Fetch linked issues.
Parse all linked issue keys from the output above. Fetch each one, prioritizing:
- Parent issues (blocks, is blocked by, parent link)
- RFE-type issues (Feature Request, RFE, Requirement)
- Related issues
Cap at 10 linked issues. For each:
Run `acli jira workitem view <KEY> --json` (or appropriate acli command) and parse the JSON output directly.
Step 3: Synthesize context.
First, determine the artifact type for each piece of work using references/artifact-hierarchy.md:
- Is this customer-facing (new capability customers will use directly)? → Feature
- Is this architectural, infrastructure, or internal tooling (enables Red Hat associates)? → Initiative
- Does it represent a measurable business outcome spanning multiple releases? → Outcome (rarely; confirm with PM)
For each distinct artifact, assess the 9 required elements (from references/feature-definition.md for Features; references/artifact-templates.md for Initiatives/Outcomes):
- Overview / Goal: [known / partial / unknown]
- Goals / expected outcomes: [known / partial / unknown]
- Requirements / Success Criteria: [known / partial / unknown]
- Out of Scope: [known / partial / unknown]
- Target Release: [known / partial / unknown]
- Background / Strategic Fit: [known / partial / unknown]
- Customer or Stakeholder Considerations: [known / partial / unknown]
- Documentation Considerations: [known / partial / unknown]
- Interoperability Considerations: [known / partial / unknown]
Also note:
- How many distinct artifacts should be created? (separate RFEs → separate artifacts)
- What type is each artifact? (Feature / Initiative / Outcome)
- Which JIRA project(s) should own each? (see
references/artifact-hierarchy.mdproject routing)
Phase 1.5: Duplicate & Existing Work Detection
Before drafting anything, search for existing Features and Initiatives that may already cover the work. Creating a duplicate wastes planning effort and fragments tracking.
Step 1: Search the target project for related Features/Initiatives.
Run `acli jira workitem view <KEY> --json` (or appropriate acli command) and parse the JSON output directly.
Step 2: Evaluate results.
| Finding | Action |
|---|---|
| Exact or near-exact match (same capability) | Stop — link the RFE to the existing Feature; do not create a new one |
| Partial overlap (some RFEs are covered) | Note which RFEs the existing Feature covers; only create new artifacts for the uncovered gap |
| Adjacent work (related but distinct) | Note the relationship; plan to link the new artifact after creation |
| No matches | Clear to proceed to Phase 2 |
Report your findings to the user before continuing.
Phase 2: Generate & Ask Interview Questions
Compare your synthesis against the required Feature elements. Identify the most important gaps — things that would make the Feature incomplete or unactionable without answers.
Generate questions that target those gaps. Pre-fill partial answers from context where possible. Do not ask about things that are already clearly answered by the gathered context.
Ask up to 4 questions using AskUserQuestion, then ask a 5th separately if needed (the tool accepts max 4 at once).
Good question framing:
- "The source issue mentions X but doesn't clarify Y — what should the acceptance criteria be for Z?"
- "Should this be one Feature or two? The RFE covers both A and B, which seem separable."
- "This looks like it could be a Feature (customer-facing) or an Initiative (internal architectural work) — which is intended?"
- "Which project should own this — ROSA or XCMSTRAT? (The issue involves cross-cutting concerns.)"
- "What is the target release for this work? The source issue doesn't specify."
Poor question framing (avoid):
- Asking about things already stated clearly in the gathered context
- Generic questions that don't reflect the specific issue content
- More than one question about the same topic
Phase 3: Draft Features
Using all gathered context plus the interview answers, draft each artifact.
Apply the correct template:
- Feature → use the template from
references/feature-definition.md - Initiative → use the template from
references/artifact-templates.md - Outcome → use the template from
references/artifact-templates.md
Fill every section with real content — no placeholders. If a section is explicitly not applicable, say so briefly.
Key quality bar:
- Market Problem must answer: who is affected, what pain, why now, what if we don't solve it
- Success Criteria must include quantified targets (%, time, deal count) — not just "feature is complete"
- Planned Epics must list 3–8 concrete work streams by name
- Timeline must include milestones by quarter/release
- All body text must use Jira wiki markup (
h2.,*,||) — never Markdown (seereferences/wiki-markup.md)
Present the draft(s) to the user:
## Draft: [Artifact Title]
**Type:** [Feature / Initiative / Outcome]
**Project:** [ROSA / XCMSTRAT / HCMPE / etc.]
**Summary:** [one-line title]
[Full body in Jira wiki markup]
Ask the user to confirm or request revisions before proceeding to creation. Do not create until approved.
Phase 4: Create in JIRA
CRITICAL: Use acli jira workitem create --from-json for creation. Descriptions must be in ADF (Atlassian Document Format), not Jira wiki markup. Convert the template body content to ADF before writing the JSON payload. See references/feature-definition.md and references/artifact-templates.md for the required fields and JSON schema.
For each approved artifact:
Step 1: Create via acli
Write the issue payload to a temp JSON file (see references/feature-definition.md for the schema), then:
acli jira workitem create --from-json /tmp/artifact-payload.json --json
Note the new issue key from the JSON output.
Step 2: Link to source issue
acli jira workitem link create --out <NEW-KEY> --in <SOURCE-KEY> --type "Implements" --yes
Step 3: Link to parent Outcome (if known)
acli jira workitem edit --key <NEW-KEY> --from-json /tmp/parent-link.json --yes
Where parent-link.json sets additionalAttributes.customfield_12313140 to the Outcome key.
Step 4: Report results
## Created Artifacts
| Key | Type | Summary | Project | Link |
|-----|------|---------|---------|------|
| ROSA-456 | Feature | ... | ROSA | $JIRA_URL/jira/browse/ROSA-456 |
| ROSA-457 | Initiative | ... | ROSA | $JIRA_URL/jira/browse/ROSA-457 |
Links created:
- ROSA-456 Implements OCPSTRAT-2666
- ROSA-457 Implements OCPSTRAT-2666
Error Handling
- JIRA_API_TOKEN or JIRA_USER not set: Tell the user: "Set
export JIRA_API_TOKEN=<your-token>andexport JIRA_USER=<your-email>before running, or run/rfe:init." - Project routing unclear: Ask the user which project to use rather than guessing.
- acli error / field errors: Show the error and ask the user whether to retry with modified fields or proceed manually.
- Issue not found: Report the key that failed and continue with the others.