Doc Plan

doc-plan

Produce a documentation plan for a strategic initiative (STRAT) or epic by analyzing all child tickets for documentation impact.

Parse arguments

$ARGUMENTS contains a Jira ticket key, typically a STRAT-level or epic-level ticket (e.g., RHAISTRAT-1401 or RHOAIENG-55000).

Step 1: Resolve the root ticket

Call the MCP tool to resolve the ticket:

mcp__mcp-atlassian__jira_get_issue(issue_key="")

Extract:

• Summary, description, status
• Issue type (Initiative, Epic, Story)
• Fix versions
• Child tickets / linked tickets

Step 2: Traverse child hierarchy

Resolve all child tickets recursively (up to 3 levels deep):

1. STRAT/Initiative → child Epics
2. Epic → child Stories/Tasks
3. Story → sub-tasks (if any)

For each ticket, resolve via MCP and extract:

• Key, summary, description, status, issue type
• Components
• Fix versions
• Labels

Cycle detection: maintain a visited set of issue keys during traversal. Before descending into any child ticket, check whether its key is already in the visited set. If so, skip that branch and record that a cycle was detected.

Limits:

• Depth cap: 3 levels maximum
• Global cap: resolve up to 100 child tickets. If more exist, note the truncation.
• If either cap is reached, stop further traversal and note the truncation in the output.

Step 3: Assess documentation impact

For each resolved child ticket, determine:

1. Does this ticket affect documentation?

• Read the summary and description
• Check if it introduces user-visible changes
• Check components (documentation-relevant components vs internal)
• Look for keywords: "API", "config", "parameter", "UI", "workflow", "deprecate"

1. What type of documentation is needed?

• Map to: newconcept, newprocedure, newreference, updateexisting, releasenote, deprecationnotice, none

1. What priority?

• critical: mentioned in acceptance criteria, GA-blocking
• high: significant user-facing change
• medium: improvement, enhancement
• low: minor change, edge case

1. Dependencies?

• Which tickets share the same feature area?
• Which tickets must be documented before this one?

Step 4: Group and organize

Group documentation items by:

• Epic (parent grouping)
• Component (cross-cutting grouping)
• Priority (work ordering)

Identify:

• Tickets that can share documentation (combined into one doc)
• Tickets that need SME input
• Tickets where implementation is not yet complete (docs should wait)

Step 5: Generate doc plan

Read ${CLAUDE_SKILL_DIR}/prompts/doc-plan.md for the planning prompt.

Construct the plan combining:

• All ticket assessments
• Grouping and dependency analysis
• Coverage summary

Step 6: Write output

Write the doc plan as markdown to stdout (for the caller to review).

Also write a structured version to workspace/doc-plan.json with the JSON format from the prompt template.

The markdown plan should include:

# Documentation Plan: 

**STRAT**:  — 
**Generated**: 
**Tickets analyzed**: 

## Summary

- **Tickets with doc impact**: N
- **Tickets without doc impact**: N
- **New concepts**: N
- **New procedures**: N
- **New references**: N
- **Updates to existing**: N
- **Release notes**: N

## Critical Priority

| Ticket | Summary | Doc Type | Modules | Dependencies |
|--------|---------|----------|---------|--------------|
| RHOAIENG-55490 | Model serving | new_procedure | proc_configure-model-serving | RHOAIENG-55489 |

## High Priority

...

## Medium Priority

...

## Low Priority

...

## No Documentation Impact

| Ticket | Summary | Reason |
|--------|---------|--------|
| RHOAIENG-55499 | Internal refactoring | No user-visible changes |

## Release Notes

| Ticket | Type | Summary |
|--------|------|---------|
| RHOAIENG-55490 | New feature | Model serving now supports custom runtimes |

Output

Primary: Markdown doc plan (displayed to caller) Secondary: workspace/doc-plan.json

Stop conditions

• Halt: Root ticket not found
• Halt: MCP Jira tools not available
• Warn: Some child tickets fail to resolve (skip and note)
• Warn: Ticket hierarchy exceeds 100 tickets (truncate and note)

Gotchas

• Traversal is capped at 3 levels deep and 100 tickets total — larger initiatives will be silently truncated, so check the output for truncation warnings.
• Circular ticket links (e.g., mutual parent-child references) are detected but can cause confusing "cycle detected" notes if Jira data is messy.
• MCP Jira tools must be available; without them the skill halts immediately with no partial output.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: opendatahub-io
• Source: opendatahub-io/ai-helpers
• License: Apache-2.0
• Homepage: https://opendatahub-io.github.io/ai-helpers/

Install and usage instructions live in the source repository linked above.