jira-issues-review

name

jira-issues-review

description

Manually triggered skill that analyzes recently resolved Jira issues using Atlassian MCP tools, classifies them for playbook relevance, and auto-invokes packmind-update-playbook with structured findings. Use when wanting to mine resolved Jira issues for recurring patterns, conventions, decisions, or action items that should be captured in the Packmind playbook. Supports non-interactive mode via `CI=true` or `--non-interactive` for scheduled runs.

Jira Issues Review Mining

Analyze recently resolved Jira issues from a configurable time window, classify them for playbook relevance, and feed actionable findings into the packmind-update-playbook workflow.

Workflow: Phase 0 → 1 → 2 → 3 → 4 → 5. Follow every phase in order.

Prerequisite: The Atlassian Rovo MCP server must be available. Load references/jira-mcp-tools.md for tool signatures and parameters.

Phase 0: Collect Parameters

Gather the required inputs before fetching any data.

Execution Mode Detection

Determine the execution mode before collecting any parameters:

1. Check CI env var: run echo $CI. If the value is true, set non_interactive = true.
2. Check skill arguments: if the skill was invoked with --non-interactive, set non_interactive = true.
3. Parse optional arguments:
   • --days N — override the default time period (default: 7 days, max: 90 days)
   • --project PROJECT_KEY — restrict analysis to a specific Jira project

If neither condition is met, non_interactive = false (interactive mode — original behavior).

Collect Parameters

1. Gather project:

   • Interactive: Ask the user which project to analyze. If they don't specify one, use mcp__atlassian__getVisibleJiraProjects to list available projects and let the user choose. If the user wants to analyze all projects, omit the project filter from JQL.
   • Non-interactive: Use --project value if provided. If not provided, analyze all projects (no project filter in JQL).

2. Gather time_period:

   • Interactive: Ask the user how far back to look (default: 7 days, max: 90 days).
   • Non-interactive: Use --days value if provided, otherwise default to 7 days silently.

   Compute the cutoff date as today - time_period in YYYY-MM-DD format.

3. Confirm before proceeding:

   • Interactive: Display and BLOCK until the user confirms:

     "Analyzing resolved Jira issues [in project KEY] for the last N days (since YYYY-MM-DD). Proceed?"

   • Non-interactive: Log the parameters and proceed automatically:

     "Non-interactive mode: analyzing resolved Jira issues [in project KEY] for the last N days (since YYYY-MM-DD)."

Phase 1: Fetch Resolved Issues

Use mcp__atlassian__searchJiraIssuesUsingJql with JQL:

project = KEY AND resolved >= -Nd ORDER BY resolved DESC

Or without project filter:

resolved >= -Nd ORDER BY resolved DESC

• Paginate using startAt and maxResults to collect all matching issues.
• Cap at 200 issues maximum. If more exist, log a warning and process only the first 200.
• If no issues are found, inform the user and stop.
• Display progress: "Found N resolved issues to analyze."

Phase 2: Fetch Details & Filter

For each issue, use mcp__atlassian__getJiraIssue to fetch the full description and comments.

Filtering

Apply these filters to discard noise:

Low-value filtering — discard issues that:

• Have no description AND no comments
• Have a description shorter than 30 characters AND no comments
• Are sub-tasks (issuetype.name == "Sub-task") with no substantial description or comments (less than 30 chars combined)
• Are bot-created and auto-resolved issues with no human comments (check if reporter is a bot/automation account and no human comments exist)

Duplicate deduplication — for issues linked as duplicates (via issuelinks with type "Duplicate"), keep only the primary issue (the one that is not marked as duplicate of another). Discard the duplicated issues.

Display progress: "Retained N substantive issues (after filtering)."

If no issues remain after filtering, inform the user and stop.

Phase 3: Classify Issues

Categorize each remaining issue by playbook relevance:

High relevance (keep)

• convention — Establishes or references a naming, structure, or organizational pattern (e.g., "All API endpoints must validate with Zod")
• decision — Records an architectural or technical decision (e.g., "Decided to use event sourcing for orders")
• best-practice — Recommends or discusses a better approach for correctness, performance, or maintainability (e.g., "Always add migration scripts for schema changes")
• recurring-pattern — Same type of issue or fix appears across multiple issues
• action-item — Commits to a specific change that affects team workflows or coding practices

Low relevance (discard)

• routine-bugfix — Standard bug fix with no generalizable lesson
• maintenance — Dependency updates, version bumps, or routine housekeeping
• one-off — Unique, non-recurring issue that doesn't generalize
• external-dependency — Issue caused by external service or third-party, no internal learning

Retain only High-relevance issues.

Recurring Theme Detection

Group retained issues by semantic similarity. A recurring theme is 2+ issues that address the same underlying pattern. For each theme:

• Assign a short descriptive label
• List all contributing issues
• Note the occurrence count

Codebase Relevance Gate

After classification, apply a strict codebase relevance filter to all high-relevance findings.

Litmus test: "Would an AI coding agent need to know this when writing, reviewing, or shipping code in this repository?" — if no, discard the signal.

Signal	Verdict	Why
"All API endpoints must validate with Zod"	KEEP	Coding convention
"Decided event sourcing for orders"	KEEP	Architecture decision
"Always add migration scripts"	KEEP	Dev workflow
"Run integration tests before merge"	KEEP	Affects how agent ships code
"Sprint velocity was 42 points"	DISCARD	Project metrics
"Move ticket to QA column after dev"	DISCARD	Process tooling
"Use Jira instead of Linear"	DISCARD	Project management tooling
"Retro action: more pair programming"	DISCARD	Team ritual

Edge cases: When a signal straddles both worlds (e.g., "label PRs with Jira ticket number"), apply the litmus test strictly — does it change how an AI agent writes or ships code? If not, discard.

Display progress: "Retained N codebase-relevant findings (discarded M non-relevant signals)."

Phase 4: Build Findings Report

Choose the output path based on execution mode:

• Interactive: .claude/tmp/jira-review-findings.md
• Non-interactive: .claude/reports/jira-review-findings-YYYY-MM-DD.md (using today's date, timestamped for CI artifact upload)

Create the target directory if needed.

Report Structure

# Jira Issues Review Findings Report

**Project**: KEY (or "All projects")
**Period**: YYYY-MM-DD to YYYY-MM-DD
**Issues analyzed**: N (after filtering)
**Actionable findings**: N
**Codebase-relevant findings**: N (of M high-relevance signals)

---

## Recurring Themes

### Theme: <label>
**Occurrences**: N issues

| Issue | Type | Reporter | Resolution Date | Summary |
|-------|------|----------|-----------------|---------|
| [PROJ-123](url) | Story | @reporter | YYYY-MM-DD | Summary of issue |
| [PROJ-456](url) | Bug | @reporter | YYYY-MM-DD | Summary of issue |

**Suggested playbook action**: <Create standard | Update standard X | Create skill | Create command | ...>
**Rationale**: <Why this theme warrants a playbook change>

---

## Individual Findings

### Finding: <short description>
- **Issue**: [PROJ-123](url)
- **Type**: Story | Bug | Task | Epic
- : @reporter
 : YYYY-MM-DD
 : label1, label2
 : Condensed description of the issue and its resolution
 : Relevant excerpts from comments
 : convention | decision | best-practice | recurring-pattern | action-item
 : 

---





| Signal summary | Reason discarded |
|---------------|-----------------|
|  |  |

URL Requirements

Every issue reference in the report MUST include the full, navigable Jira URL (e.g., https://team.atlassian.net/browse/PROJ-123). This is critical because packmind-update-playbook extracts source URLs from the findings report for its packmind-cli diff --submit -m source attribution. The expected attribution format is:

<topic>: <summary> (source: Jira PROJ-123 https://team.atlassian.net/browse/PROJ-123)

Construct the URL from the API response: use the self field's base URL or build https://<site>.atlassian.net/browse/<issueKey> from the issue key. Never use placeholder (url) — always substitute the real URL.

Phase 5: Present and Hand Off

Interactive mode

1. Display a summary to the user:

   • Number of recurring themes found
   • Number of individual findings
   • List of suggested playbook actions

2. Ask the user:

   "Found N recurring themes and M individual findings. Review the full report at .claude/tmp/jira-review-findings.md. Proceed to update the playbook with these findings?"

3. On confirm: Invoke the packmind-update-playbook skill with the full report content as the intent. This maps to Case B (explicit intent) of that skill's Phase 0. Frame the intent as:

   "Update the Packmind playbook based on the following Jira issues review findings report: "

4. On decline: Inform the user the report is available at .claude/tmp/jira-review-findings.md for manual review.

Non-interactive mode

1. Display a summary (same as interactive: themes count, findings count, suggested actions).

2. Log the report output path:

   "Report written to .claude/reports/jira-review-findings-YYYY-MM-DD.md."

3. Invoke packmind-update-playbook with the full report content as the intent, passing along non-interactive mode. Frame the intent as:

   "Update the Packmind playbook based on the following Jira issues review findings report: "

   The packmind-update-playbook skill will auto-approve all changes and delegate creation of new artifacts to the corresponding creation skills (packmind-create-standard, packmind-create-command, packmind-create-skill), all running in non-interactive mode.

Resources

references/

• jira-mcp-tools.md — Atlassian Jira MCP tool signatures and parameters used by this skill