agentleFS
ehnruizehnruiz/claude-skills/skills/epic-review/SKILL.md

epic-review

Skill3 starsChanged 31 days agoMITView on GitHub

Review a Jira epic (or the Jira Product Discovery idea it implements) against its own In Scope list - which items are done, partly done, not started, or have no evidence - then flag scope conflicts and stale statuses, and draft the In Scope list for the follow-on epic. Use this whenever the user is preparing for an epic review, quarter close-out, scope or refinement meeting, asks "what was completed on this epic", "what is left for Q2", "what should carry over", "prep me for the review of PROJ-1234", pastes a JPD idea or epic link with a meeting invite, or needs to fill in a cloned epic whose scope still says TBD.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    JPD idea: follow is implemented by links to the delivery epic(s).

    line 38
  2. Very specific:

    Epic: follow implements back to the idea, if any, and is cloned by forward to the next-period epic.

    line 40
  3. Distinctive:

    Do not fetch every ticket's description.

    line 61
  4. Distinctive:

    Carry over: every ❌ and 🟡 item from step 3, with keys.

    line 96

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Epic review

Answer the question a quarter-close meeting asks: of what this epic promised, what is done, what is left, and what goes into the next one. The answer comes from ticket evidence, item by item, not from the epic's status field, which is usually stale.

Requires: the Atlassian MCP connector (getJiraIssue, searchJiraIssuesUsingJql, searchConfluence, and listJiraIssueComments through executeRead). Read-only by default. Companion skills: ticket-write files any missing work this review uncovers, ticket-digest recaps it.

Team defaults

The rules below are team-neutral. These values are examples. Read the real ones from the user's CLAUDE.md or from the epic itself, and ask once when neither says:

SettingExample
Jira site / delivery projectyour-site.atlassian.net, project PROJ
Discovery (JPD) projectDISC; ideas link to delivery epics with the Polaris work item link type (implements / is implemented by)
Epic summary shapeT1 | Q1FY27 | Area | Name (Phase n - ...); the next quarter's epic is usually a clone, linked with the Cloners type
Scope sections in descriptions### In Scope and ### Out of Scope under ## Requirements
Docs homeConfluence, the team's space

Workflow

1. Resolve the anchor

The user may give a JPD idea URL (.../polaris/projects/KEY/ideas/...?selectedIssue=KEY-13), an epic key, or a board link plus a meeting invite. Take the key from selectedIssue when it is there. Fetch it with getJiraIssue, view: "evidence", which also returns issue links.

  • JPD idea: follow is implemented by links to the delivery epic(s). Keep the idea's roadmap and target-date fields; they say which quarter the review closes.
  • Epic: follow implements back to the idea, if any, and is cloned by forward to the next-period epic.

You now have up to three records: idea, epic, next epic.

2. Collect

  1. The epic's description, read for its In Scope and Out of Scope bullets. If the idea has its own lists, read those too; step 4 compares them.
  2. Epic comments: executeRead with name: "listJiraIssueComments". Earlier rounds of this same question often sit there ("which of these were not completed?"), with an answer from the epic owner.
  3. Children of each epic, one query per epic: parent = PROJ-1200, fields summary, status, assignee. Query them separately; a single parent in (...) query does not reliably say which epic each row belongs to.
  4. One sweep for related tickets outside the epics: the project's area prefix or label in the summary (summary ~ "CHECKOUT", labels = checkout). Decision, assumption and open-question registers often live there as sub-tasks of an unparented task.
  5. Optional: searchConfluence in the docs home for pages titled with the area name, for step 5's documentation question. Titles and last-modified dates are enough.

Do not fetch every ticket's description. Summaries and statuses carry the evidence.

3. Map each In Scope item to evidence

One row per In Scope bullet, in the epic's own order and wording (shortened if long). For each, pick the tickets that deliver it and give one status:

MarkMeaning
✅ DoneEvery delivering ticket is Done
🟡 PartialSome done, or the rest are In Review / Working
❌ Not startedDelivering tickets exist and are all To Do / Backlog, or the epic owner says so in a comment
❓ No evidenceNo ticket, comment or page shows this item; say so, do not guess

A ticket already parented to the next epic counts as moved; say "moved to Q2" in the evidence cell. Evidence is ticket keys plus status, under 20 words. Name the assignee only when it matters (unassigned work on the critical path, or the owner is in the meeting).

4. Flag what the meeting must decide

At most five flags, most consequential first. Look for:

  • Scope conflicts: the idea and the epic list the same item on different sides, or an item appears in both In and Out of Scope.
  • Status drift: the epic says To Do while most children are Done, or the idea's target date has passed.
  • Unowned critical work: the one remaining item is unassigned or in Backlog.
  • Open policy: assumptions that contradict each other, or business decisions still open that the next phase depends on.
  • Empty successor: the next epic's scope still reads TBD.

5. Draft the follow-on scope

A paste-ready In Scope list for the next epic, grouped:

  1. Carry over: every ❌ and 🟡 item from step 3, with keys.
  2. Already parented to the next epic, with keys.
  3. Candidates the user's own role or the sweep in step 2 suggests. Mark these as proposals.

Then an Out of Scope list, starting from the current epic's. If the documentation home is part of the outcomes, name the existing pages from step 2.5 and propose a hub structure in one line per child page.

Output contract

In this order, in chat:

  1. One sentence naming the idea, epic and next epic with keys, and what the meeting must produce.
  2. The step 3 table: In Scope item | Status | Evidence.
  3. The step 4 flags, numbered.
  4. The step 5 draft.
  5. One line offering the write-back (below).

Budget: the whole output under 600 words, table included. Plain language; a ticket key is the only identifier allowed outside the evidence column.

Example table rows (placeholder keys):

| Configure CI/CD pipelines | ❌ Not started | PROJ-1205 PR env Backlog; PROJ-1206 staging Backlog, unassigned, moved to Q2 |
| Locale routing foundation | ✅ Done | PROJ-1203 middleware Done; PROJ-1190 URL decision Done |
| Present to leadership to finalize | ❓ No evidence | No ticket or notes page |

Writing back

This skill does not change Jira on its own. Writing the draft into the next epic happens only when the user asks, and then:

  1. getJiraIssue the next epic with responseContentFormat: "html" and keep its current description.
  2. Replace only the TBD bullets under In Scope and Out of Scope, leaving Background and Stakeholders untouched.
  3. editJiraIssue with contentFormat: "html" so links and mentions survive.
  4. Quote the before and after to the user.

New work the review uncovered goes through ticket-write, one ticket at a time, after the user confirms the list.

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.