agentleFS
jcesarperezjcesarperez/claude-em/.claude/skills/check-issue/SKILL.md

check-issue

Skill94 starsChanged 5 months agoMITView on GitHub

Evaluate whether a Jira issue (Story, Bug, Task) is stuck. Analyzes execution health (time in status, ping pong, PR state, subtask movement), with lighter definition and decomposition checks adapted per issue type. Use whenever the user wants to know if an issue is blocked, stuck, or at risk.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

24 rules. About 0 are common, 14 are distinctive, and 10 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈14

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 10

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Run this script using the Bash tool:

    line 12
  2. Distinctive:

    If the issue is not found, the script exits with an error.

    line 19
  3. Very specific:

    If the argument matches [A-Z]+-\d+, use it directly.

    line 90
  4. Distinctive:

    Use the jira skill, then CLI, then REST API.

    line 93
  5. Distinctive:

    If multiple matches, list them and ask the user to confirm.

    line 94
  6. Distinctive:

    Run the script; stop and report error if it fails

    line 98
  7. Distinctive:

    Run the three health analyses using the fetched data

    line 99
  8. Distinctive:

    No subtasks but description implies multiple distinct pieces of work → ⚠️ may be oversized

    line 146
  9. Distinctive:

    All subtasks Done but issue still open → inconsistency, flag it

    line 148
  10. Very specific:

    Use issue.status (exact name, not category) and issue.time_in_current_status.seconds:

    line 161
  11. Very specific:

    Only apply this to issues in statusCategory == "In Progress".

    line 168
  12. Very specific:

    Use pingpong pre-computed object AND read issue.comments directly to assess context.

    line 172
  13. Very specific:

    Read issue.comments and look for:

    line 180
  14. Very specific:

    Use issue.pr and issue.pr.items (sorted by last_updated desc).

    line 195
  15. Very specific:

    Use issue.cycle_time.seconds (Done issues) or issue.time_in_progress.seconds (In Progress issues).

    line 212
  16. Very specific:

    Use issue.pr_merge_gap.seconds.

    line 222
  17. Distinctive:

    All subtasks Done but parent issue not Done → inconsistency

    line 234
  18. Very specific:

    No subtasks started (statusCategory == "To Do" for all) → nothing moving

    line 235
  19. Distinctive:

    If definition is empty, state it explicitly.

    line 309
  20. Distinctive:

    DO NOT invent data not present in the input

    line 314
  21. Distinctive:

    DO NOT flag Decomposition for Sub-tasks

    line 315
  22. Distinctive:

    DO NOT flag unassigned issues that are in "To Do" status category — only flag unassigned in "In Progress"

    line 316
  23. Distinctive:

    DO NOT apply stuck thresholds to issues in "Done" or "To Do" status categories

    line 317
  24. Very specific:

    If pr.last_updated is null, omit the staleness assessment for PRs

    line 318

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Tool

Run this script using the Bash tool:

bash .claude/skills/check-issue/scripts/fetch_issue.sh <ISSUE_KEY> --json
  • ISSUE_KEY: any issue key (Story, Bug, Task, Sub-task)
  • If the issue is not found, the script exits with an error. Report it and stop.

The script outputs JSON with this structure:

{
  "fetched_at": "2026-04-20T10:00:00Z",
  "issue": {
    "key": "FBX-1234",
    "issuetype": "Story",
    "summary": "...",
    "status": "In Review",
    "statusCategory": "In Progress",
    "description": "plain text",
    "assignee": "Name or null",
    "priority": "High",
    "duedate": "YYYY-MM-DD or null",
    "created": "...",
    "updated": "...",
    "comments": [{ "author": "Name", "created": "...", "body": "plain text" }],
    "time_in_current_status": { "formatted": "Xd Yh Zm", "seconds": 0, "since": "..." },
    "first_in_progress_at": "YYYY-MM-DDTHH:MM:SSZ or null",
    "cycle_time": { "formatted": "Xd Yh Zm", "seconds": 0 }, // Done issues only, or null
    "time_in_progress": { "formatted": "Xd Yh Zm", "seconds": 0 }, // In Progress issues only, or null
    "status_transitions": [{ "from": "...", "to": "...", "at": "...", "author": "Name" }],
    "assignee_changes": [{ "from": "...", "to": "...", "at": "...", "changed_by": "Name" }],
    "pr": {
      "count": 2,
      "open_count": 1,
      "draft_count": 0,
      "items": [                              // null if only aggregate data is available
        {
          "id": "#123",
          "title": "fix: FBX-1234 ...",
          "status": "OPEN",                   // OPEN | DRAFT | MERGED | DECLINED | CLOSED
          "open": true,                       // true only when status == OPEN (ready for review)
          "author": "Name",
          "branch": "FBX-1234-my-branch",
          "last_updated": "YYYY-MM-DDTHH:MM:SS.000+0000",
          "url": "https://github.com/..."
        }
      ]
    } // or null if no PRs found
    "pr_merge_gap": { "formatted": "Xd Yh Zm", "seconds": 0 } // gap between first and last MERGED PR; null if < 2 merged PRs
  },
  "parent_epic": { "key": "FBX-1000", "summary": "...", "status": "In Development" }, // or null
  "subtasks": [
    {
      "key": "FBX-1235",
      "summary": "...",
      "status": "In Progress",
      "statusCategory": "In Progress",
      "assignee": "Name or null",
      "time_in_current_status": { "formatted": "Xd Yh Zm", "seconds": 0, "since": "..." }
    }
  ],
  "pingpong": {
    "status_bounces": 0,
    "assignee_change_count": 0,
    "pingpong_comment_count": 0,
    "comment_author_count": 0,
    "detected": false
  }
}

Instructions

Issue Key Resolution

  • If the argument matches [A-Z]+-\d+, use it directly.
  • Otherwise, treat it as a search phrase and search Jira using JQL: summary ~ "<phrase>" ORDER BY updated DESC Use the jira skill, then CLI, then REST API.
  • If multiple matches, list them and ask the user to confirm.

Analysis Steps

  1. Run the script; stop and report error if it fails
  2. Run the three health analyses using the fetched data
  3. Produce the structured report

Scoring Rubrics

Definition Health (lightweight, type-specific)

Evaluate issue.description and issue.comments. Assess each criterion for the issue type:

Story:

#Criterion✅
1User value / goalClear description of what the user can do or what changes
2Acceptance criteriaExplicit AC, "done when", or testable conditions present
3Dependencies flaggedNamed blockers, teams, or linked issues if applicable

Bug:

#Criterion✅
1Steps to reproduceClear steps or scenario that triggers the bug
2Expected vs actualBoth states described
3Severity/priority setPriority field is not null/default

Task / Sub-task:

#Criterion✅
1Clear goalDescription states what needs to be done
2Scope definedWhat is in scope; not ambiguous

Score:

  • 🟢 All ✅ or at most 1 ⚠️
  • 🟡 1–2 ⚠️, no ❌
  • 🔴 Any ❌, or description is empty

Decomposition Health (lightweight)

Evaluate subtasks array. Skip this section entirely if issuetype == "Sub-task".

#Criterion✅
1Appropriate decompositionComplex issue has subtasks; simple issue may not need them
2Not oversizedsubtasks.length ≤ 5; no single subtask has been stuck significantly longer than others
3All subtasks have clear summariesSummaries are descriptive, not placeholders

Signals to flag:

  • No subtasks but description implies multiple distinct pieces of work → ⚠️ may be oversized
  • subtasks.length > 5 → ⚠️ may be over-decomposed or should be an epic
  • All subtasks Done but issue still open → inconsistency, flag it

Score:

  • 🟢 All ✅ or at most 1 ⚠️
  • 🟡 1 ⚠️
  • 🔴 Any ❌

Execution Health (primary focus)

This is the core of the analysis. Evaluate all signals, then derive overall state.

Status Stuck Signal

Use issue.status (exact name, not category) and issue.time_in_current_status.seconds:

StatusStuck thresholdCritical threshold
Statuses containing "Review", "QA", "Testing"3 days (259200s)5 days
All other In Progress category statuses5 days (432000s)10 days

Only apply this to issues in statusCategory == "In Progress". Ignore To Do and Done.

Ping Pong (primary signal for Bugs, applies to all types)

Use pingpong pre-computed object AND read issue.comments directly to assess context.

Pre-computed signals:

  • status_bounces ≥ 2: issue moved back and forth between in-progress category statuses
  • assignee_change_count ≥ 3: passed between people multiple times
  • pingpong_comment_count ≥ 3 AND comment_author_count ≥ 2: back-and-forth in comments

Manual comment analysis (always do this): Read issue.comments and look for:

  • Open questions directed at another person ("@name can you...", "waiting for X to...")
  • Unanswered questions (question asked, no reply author after it, or reply is another question)
  • Comments that pass the ball ("handing back to", "assigned to you", "please check", "any update?")
  • Long gaps between comment threads that suggest waiting

Report specific evidence: quote or paraphrase the comment and its author/date.

Ping pong severity:

  • 🔴 Critical: pingpong.detected == true OR clear evidence of unanswered blocking question older than 3 days
  • 🟡 At Risk: 1–2 signals, or recent back-and-forth with open question
  • 🟢 None: no signals

PR Health

Use issue.pr and issue.pr.items (sorted by last_updated desc). Read the PR list to understand what happened: how many attempts, rework, drafts that never progressed, etc.

SituationSignal
status contains "Review" AND pr == null⚠️ In review but no PR — status may be wrong
pr.open_count > 0 AND most recent open PR last_updated > 3 days ago🟡 PR open but stale
pr.open_count > 0 AND most recent open PR last_updated > 7 days ago🔴 PR open and very stale
pr.draft_count > 0 AND statusCategory == "In Progress" AND most recent draft last_updated > 5 days ago🟡 Draft stale — engineer may be blocked
pr.draft_count > 0 AND all other PRs are MERGEDℹ️ Drafts never promoted — note but do not flag
all PRs MERGED AND statusCategory != "Done"⚠️ All PRs merged but issue still open — status stale
pr.count > 2 AND multiple DECLINED or many attempts⚠️ Rework signal — investigate why
statusCategory == "In Progress" AND pr == nullℹ️ No PR yet — expected if early in-progress

Skip PR analysis if pr == null and status is not review-related.

Cycle Time Signal

Use issue.cycle_time.seconds (Done issues) or issue.time_in_progress.seconds (In Progress issues). Apply to both Done and In Progress; ignore To Do.

ElapsedSignal
> 30 days (2,592,000s)🔴 Very long cycle time — investigate delivery pace
> 15 days (1,296,000s)🟡 Getting long — monitor
≤ 15 days🟢 OK

PR Merge Gap Signal

Use issue.pr_merge_gap.seconds. Skip if pr_merge_gap == null (fewer than 2 merged PRs).

Gap between first and last MERGED PRSignal
> 14 days (1,209,600s)🔴 Large gap — delivery stretched, likely paused or blocked mid-way
> 7 days (604,800s)🟡 Noticeable gap — work spread across multiple cycles
≤ 7 days🟢 OK

Subtask Execution

Use subtasks array:

  • Any subtask stuck by the same thresholds → flag with key + time
  • All subtasks Done but parent issue not Done → inconsistency
  • No subtasks started (statusCategory == "To Do" for all) → nothing moving

Assignee

  • assignee == null AND statusCategory == "In Progress" → 🔴 active work with no owner

Overall Execution State

  • 🟢 Moving: within stuck threshold, no ping pong, PR healthy (or not applicable), subtasks progressing, cycle time ≤ 15d, PR merge gap ≤ 7d
  • 🟡 At Risk: approaching threshold (>50% of limit), minor ping pong signs, PR open but not stale, 1 subtask stuck, cycle time > 15d, OR PR merge gap > 7d
  • 🔴 Stuck: exceeded stuck threshold, OR pingpong.detected == true OR clear unanswered blocking question, OR assignee == null in progress, OR PR very stale, OR all subtasks blocked, OR cycle time > 30d, OR PR merge gap > 14d

Output Format

Return ONLY this structure:


🎫 Issue Snapshot — KEY: Summary

  • Type: Story / Bug / Task / Sub-task
  • Status: "Status name" [statusCategory] — in this status for Xd Yh Zm (since YYYY-MM-DD)
  • Progress: First in progress YYYY-MM-DD · Cycle time Xd Yh (Done) | In progress Xd Yh total (In Progress) | — (To Do / never started)
  • Assignee: Name | ⚠️ Unassigned
  • Priority: High / Medium / Low / —
  • PRs: 3 total · 1 OPEN · 1 DRAFT · 1 MERGED (last activity YYYY-MM-DD) | No PRs (list each PR on its own line if ≤ 4 total: " · #123 MERGED — branch-name — author (YYYY-MM-DD)")
  • Parent: KEY — Epic Summary [Epic Status] | No parent
  • Subtasks: X total (X Done · X In Progress · X To Do) | None
  • Comments: X

🧠 Definition Health — 🟢 / 🟡 / 🔴

(Use type-specific criteria from the rubric)

  • ✅/⚠️/❌ Criterion: one-line assessment

🧩 Decomposition Health — 🟢 / 🟡 / 🔴 | N/A (Sub-task)

  • ✅/⚠️/❌ Criterion: one-line assessment
  • (Flag inconsistencies: all subtasks Done but issue open, etc.)

⚙️ Execution Health — 🟢 Moving / 🟡 At Risk / 🔴 Stuck

  • Status time: Xd in "Status" — 🟢 OK / 🟡 Approaching / 🔴 Stuck (threshold: Xd)
  • Ping pong: 🔴 Detected / 🟡 Signals present / 🟢 None
    • Evidence: "[quote or paraphrase]" — Author (date) | None
  • Cycle time: Xd total — 🟢 OK / 🟡 Long (>15d) / 🔴 Very long (>30d) | — (To Do)
  • PR merge gap: Xd between first and last merge — 🟢 OK / 🟡 Spread (>7d) / 🔴 Large (>14d) | N/A
  • PR: state — assessment | No PR
  • Subtasks: X moving / KEY stuck Xd / all Done but issue open | None
  • Assignee: ✅ Assigned / 🔴 Unassigned while In Progress
  • Key risks: bullet list

🧭 Summary

Overall: 🟢/🟡/🔴 · Definition: 🟢/🟡/🔴 · Decomposition: 🟢/🟡/🔴 · Execution: 🟢/🟡/🔴

  • 3–5 bullets: current state · main risk · main blocker

🎯 Recommendations

  1. [Verb] Specific concrete action
  2. ... (2–5 items, ordered by urgency)

Style Guidelines

  • Cite status names, durations, dates, and comment authors — no vague observations
  • For ping pong: quote or paraphrase actual evidence from comments; never say "ping pong detected" without backing it up
  • Bugs: pay extra attention to ping pong and blocking questions — this is the most common reason bugs stall
  • If definition is empty, state it explicitly; don't skip the section
  • Be concise and direct — EM tone, no fluff

Constraints

  • DO NOT invent data not present in the input
  • DO NOT flag Decomposition for Sub-tasks
  • DO NOT flag unassigned issues that are in "To Do" status category — only flag unassigned in "In Progress"
  • DO NOT apply stuck thresholds to issues in "Done" or "To Do" status categories
  • If pr.last_updated is null, omit the staleness assessment for PRs

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.