agentleFS
solo-istsolo-ist/prose/.claude/skills/roadmap-refinement/SKILL.md

roadmap-refinement

Skill8 starsChanged 5 months agoMITView on GitHub

Audit open issues and project board. Flags done-but-open issues, board drift, and stale items for bulk cleanup.

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

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

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

    No arguments. Operates on solo-ist/prose and the Prose Roadmap project (project #5).

    line 20
  2. Very specific:

    Use --jq for data extraction instead of piping to external processors (e.g., python3, jq).

    line 36
  3. Distinctive:

    Wait for the user to review the report.

    line 108
  4. Very specific:

    Important: Use individual parallel Bash calls for batch operations — never for loops.

    line 110
  5. Very specific:

    Prefer item-archive over item-delete: archiving is reversible (item-unarchive) and preserves the tracked record.

    line 121
  6. Distinctive:

    If the board needs a new status column or option changes, tell the user to do it manually in the GitHub UI.

    line 154
  7. Distinctive:

    Board is source of truth for priority: If an issue isn't on the board, it's not prioritized.

    line 166
  8. Distinctive:

    Never mutate field definitions: Only move items between existing columns.

    line 170

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

Roadmap Refinement

Audit the backlog and project board, then present findings for approval before making changes.

Keep docs/roadmap.md in sync

docs/roadmap.md is the canonical roadmap narrative — what we're building, the wave model, the track structure, and the A→B→C merge ladder. The project board (#5) is the live queue (column = phase, milestone = wave). These two must stay consistent. Board cleanup that changes the narrative — an epic completes, a wave's scope shifts because issues were closed or superseded, track sequencing changes — must be reflected in docs/roadmap.md in the same pass, or the doc silently drifts out of date. Read the doc before auditing; propose doc edits alongside the board edits.

Usage

/roadmap-refinement

No arguments. Operates on solo-ist/prose and the Prose Roadmap project (project #5).

Workflow

1. Gather State

Fetch all data in parallel using gh CLI:

gh issue list --repo solo-ist/prose --state open --limit 200 --json number,title,labels,updatedAt,createdAt,comments,body,milestone
gh pr list --repo solo-ist/prose --state all --limit 200 --json number,title,state,mergedAt,body,headRefName,closedAt
gh project item-list 5 --owner solo-ist --format json --limit 500

Also read the current docs/roadmap.md — it is the narrative the board should match, so you can flag where the two have drifted. If your worktree may be behind main, read the live version with git show origin/main:docs/roadmap.md after git fetch origin.

Use --jq for data extraction instead of piping to external processors (e.g., python3, jq). This keeps each command as a single gh invocation that matches the Bash allowlist:

gh project item-list 5 --owner solo-ist --format json --limit 500 --jq '.items[] | "\(.content.number)\t\(.id)\t\(.status)\t\(.content.type)\t\(.content.title)"'

Important: gh project item-list returns ALL items regardless of issue state — the GitHub Projects UI is:open filter is view-only and not applied by the API. After fetching, build an open issues set from the issues list and use it to distinguish open vs closed board items throughout the analysis.

2. Cross-Reference: Done but Open

For each open issue, check if it's likely completed:

  • Linked merged PRs: Issue referenced in a merged PR's body or commit messages
  • Closed branches: An issue-<number>-* branch exists but the issue is open
  • Stale in-flight work: A linked open PR or issue-<number>-* branch exists but the issue has had no activity in 30+ days (may have been quietly merged or abandoned)

Flag these as candidates for closing. Do NOT close automatically.

3. Board Drift

Compare open issues against the project board:

  • Missing from board: Open issues not tracked in the project
  • No status: Items on the board with null/empty status
  • Closed but on board: Issues closed but still showing as active on the board
  • Superseded: Issues whose work has been absorbed by other issues

4. Staleness Check

Flag issues with no activity (comments, commits, PR references) in 60+ days. These may need:

  • Closing as stale
  • Re-prioritization
  • A comment to confirm still relevant

5. Present Findings

Output a structured report:

## Backlog Audit

### Ready to Close (done but still open)
| # | Title | Evidence | Suggested Note |
|---|-------|----------|----------------|
| 42 | Fix login | PR #50 merged | Done — fixed in #50 |

### Board Drift
| Issue | Problem | Action |
|-------|---------|--------|
| #55 | Missing from board | Add to [column] |
| #60 | Status is null | Set to [column] |

### Stale (60+ days inactive)
| # | Title | Last Activity | Recommendation |
|---|-------|---------------|----------------|
| 30 | Old feature | 2025-10-01 | Close or re-prioritize |

### Roadmap Doc Updates (`docs/roadmap.md`)
| Section | Drift | Proposed Edit |
|---------|-------|---------------|
| Wave 1 / Track B | #312 shipped but still listed as in-flight | Move to "shipped" / retire from active wave |

### Summary
- X issues ready to close
- X board items need status updates
- X stale issues to review
- X `docs/roadmap.md` sections need updating to match the board

6. Execute on Approval

Wait for the user to review the report. Then use only safe operations.

Important: Use individual parallel Bash calls for batch operations — never for loops. Loops are shell constructs that don't match the Bash allowlist and force manual approval on every invocation. Individual calls auto-approve and run concurrently (faster too).

Closing issues:

gh issue close <NUMBER> --repo solo-ist/prose --comment "..."

Archiving completed/closed items (the board has no Done column — archive is the terminal state):

gh project item-archive 5 --owner solo-ist --id <ITEM_ID>

Prefer item-archive over item-delete: archiving is reversible (item-unarchive) and preserves the tracked record. Reserve item-delete for items added by mistake that should leave no trace.

Moving items between columns:

gh project item-edit --project-id <PROJECT_ID> --id <ITEM_ID> --field-id <STATUS_FIELD_ID> --single-select-option-id <OPTION_ID>

Adding issues to the board:

gh project item-add 5 --owner solo-ist --url <ISSUE_URL>

Updating the roadmap narrative: When approved board changes alter the roadmap (an epic completes, a wave's scope shifts, sequencing changes), edit docs/roadmap.md in the same pass so the narrative and the board stay consistent. This is a normal file edit — commit it on a branch / open a PR; it is not a gh project operation. Don't archive a completed epic's board items without also retiring it from the doc's active waves.

Board Columns

Current status options on the Prose Roadmap board:

ColumnPurpose
User RequestedFeature requests from users — needs triage
Do FirstHighest priority — work on these now
Do NextNext up after current work completes
LaterBacklog — not yet prioritized
Quick WinsLow effort, high value — ship alongside bigger work
On HoldBlocked or paused

There is no Done/terminal column — every option above is an active-work bucket — and no "In Progress" column; track in-flight work via open PRs and issue-<number>-* branches, not a board status. When an issue is closed or its PR merges, archive its board item (gh project item-archive); do not leave it sitting in an active column, and do not delete it.

Dangerous Operations — DO NOT USE

NEVER use updateProjectV2Field to modify single-select field options. This GraphQL mutation replaces option IDs, which silently disconnects every board item from its status — effectively wiping the entire board. This has happened before and required manual recovery.

If the board needs a new status column or option changes, tell the user to do it manually in the GitHub UI.

Safe operations are limited to:

  • gh project item-edit — move items between existing columns
  • gh project item-archive — archive completed/closed items (reversible terminal state; the board has no Done column)
  • gh project item-add — add items to the board
  • gh project item-delete — remove an item entirely (use only for mistaken additions; prefer archive for done work)

Key Principles

  • Never close issues without approval. Present, don't act.
  • Evidence over assumptions. Link to the merged PR or commit that completed the work.
  • Board is source of truth for priority. If an issue isn't on the board, it's not prioritized.
  • Stale != irrelevant. Flag staleness, but let the user decide disposition.
  • Done items are archived, not deleted. There is no Done column — archive closed/merged items off the active board with item-archive (reversible). Don't leave them in an active column; don't item-delete them.
  • The doc and the board move together. docs/roadmap.md is the narrative; board #5 is the queue. Any cleanup that changes the narrative gets reflected in the doc in the same pass — never let the doc drift behind the board.
  • Never mutate field definitions. Only move items between existing columns.

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.