List PRs

Overview

Scans all open, non-draft PRs across OmniNode-ai repos and prints a grouped dashboard showing
CI status, mergeable state, and review decision. PRs are grouped into five buckets:
READY TO MERGE, FAILING, PENDING, CONFLICTS, and NO CI RUN. Use --repo to focus on one
repo or --ready-only to filter to actionable PRs only. This skill is read-only: it fetches
data via gh but makes no changes to any repo or PR.

Announce at start: "I'm using the list-prs skill to scan open PRs."

Quick Start

/list-prs
/list-prs --repo omniclaude
/list-prs --ready-only
/list-prs --repo omnibase_core --ready-only
/list-prs --include-drafts

Repos Scanned (Default)

Short name	GitHub slug
omniclaude	OmniNode-ai/omniclaude
omnibase_core	OmniNode-ai/omnibase_core
omnibase_infra	OmniNode-ai/omnibase_infra
omnibase_spi	OmniNode-ai/omnibase_spi
omniintelligence	OmniNode-ai/omniintelligence
omnimemory	OmniNode-ai/omnimemory
omnidash	OmniNode-ai/omnidash
onex_change_control	OmniNode-ai/onex_change_control

omninode_infra is excluded by default. Pass --repo omninode_infra to include it explicitly.

Execution Steps

1. Resolve repo list: If --repo is given, resolve it to the full OmniNode-ai/<repo>
   slug (accepting either the short name or the full slug). Otherwise use the default list above.

2. Fetch PR data: For each repo, run:

   gh pr list \
     --repo OmniNode-ai/<repo> \
     --state open \
     --json number,title,author,headRefName,isDraft,mergeable,mergeStateStatus,reviewDecision,statusCheckRollup,createdAt,url \
     --limit 50

   Use ONLY the exact fields listed above — do not add baseRepository, headRepository,
   or any other fields. The gh CLI available in this environment does not support all GraphQL
   fields and will fail with "Unknown JSON field" if extras are added.

   Fallback: If the gh version does not support mergeStateStatus (returns "Unknown JSON
   field"), retry the query without it. In that case, fall back to mergeable == "CONFLICTING"
   for CONFLICTS classification (the pre-v1.1.0 behavior).

   If --include-drafts is NOT set, filter out any PR where isDraft == true.

3. Classify each PR using the rules in the Classification section below.

4. If --ready-only: discard all PRs except those in the READY TO MERGE bucket.

5. Fetch changed files for CONFLICTS PRs: For each PR in the CONFLICTS bucket (capped
   at the first 10 PRs), fetch changed files:

   gh pr diff --repo OmniNode-ai/<repo> --name-only <pr_number>

   Store the first 8 file paths per PR. If more than 8 files are changed, note the overflow
   count (e.g., ... and 4 more files). If gh pr diff fails, skip file listing for that PR
   and continue.

6. Print output using the format in the Output Format section below.

7. Print summary line at the end with totals per bucket.

PR Classification

Assign each PR to exactly one bucket (checked in this order):

Bucket	Condition
READY TO MERGE	mergeable == "MERGEABLE" AND all required checks passed AND reviewDecision in ("APPROVED", null, "")
CONFLICTS	mergeStateStatus == "DIRTY" (actual merge conflicts). If mergeStateStatus is unavailable (fallback mode), use mergeable == "CONFLICTING" instead. This distinguishes real file conflicts from PRs that are merely blocked by reviews or policies.
FAILING	Not CONFLICTS AND any required check has conclusion == "FAILURE" or conclusion == "TIMED_OUT"
PENDING	Not CONFLICTS AND at least one required check is in progress or queued (no failures)
NO CI RUN	statusCheckRollup is empty or null AND not CONFLICTS

CI status helpers (apply to required checks only — checks where isRequired == true):

def required_checks(pr):
    return [c for c in (pr.get("statusCheckRollup") or []) if c.get("isRequired", False)]

def ci_status(pr):
    checks = required_checks(pr)
    if not checks:
        return "no_ci"
    if any(c.get("conclusion") in ("FAILURE", "TIMED_OUT") for c in checks):
        return "failing"
    if any(c.get("status") in ("IN_PROGRESS", "QUEUED", "WAITING", "PENDING") for c in checks):
        return "pending"
    if all(c.get("conclusion") == "SUCCESS" for c in checks):
        return "green"
    return "no_ci"

If mergeable == "UNKNOWN", treat as PENDING (GitHub is still computing mergeability).

Output Format

=== Open PRs — OmniNode-ai (2026-02-23) ===

--- READY TO MERGE (2) ---
  omniclaude        #247  jonah/omn-2345-fix-hook-timeout      CI: green    Approved
  omnibase_core     #183  jonah/omn-2290-pydantic-v2-upgrade   CI: green    (no review)

--- FAILING (1) ---
  omniintelligence  #91   jonah/omn-2301-drift-classifier      CI: failing  Approved
      failing checks: quality, test

--- PENDING (3) ---
  omnibase_infra    #134  jonah/omn-2355-kafka-dlq             CI: pending  (no review)
  omnibase_infra    #133  jonah/omn-2340-session-store         CI: pending  (no review)
  omnidash          #47   jonah/omn-2360-analytics-panel       CI: pending  (no review)

--- CONFLICTS (1) ---
  omnibase_spi      #28   jonah/omn-2200-spi-refactor          CI: green    (no review)
      changed files:
        src/omnibase_spi/protocols/handler.py
        src/omnibase_spi/protocols/registry.py
        tests/unit/test_handler.py

--- NO CI RUN (1) ---
  omnimemory        #62   jonah/omn-2290-ingestion-rewrite     CI: none     (no review)

TOTAL: 8 open PRs — 2 ready, 1 failing, 3 pending, 1 conflicts, 1 no CI

Column widths:

• Repo name: left-aligned, padded to 18 chars
• PR number: right-aligned, 4 chars (#247)
• Branch name: left-aligned, truncated to 45 chars if longer
• CI badge: one of CI: green, CI: failing, CI: pending, CI: none
• Review: Approved, Changes requested, (no review), Review required

For FAILING PRs, print a second indented line listing the names of the failing checks
(comma-separated). Limit to the first 5 check names; append ... if more.

For CONFLICTS PRs, print indented lines showing changed files (fetched in step 5). Format:

• First line: changed files: (6-space indent)
• Each file: <filepath> (8-space indent)
• Show at most 8 files per PR; if more, append ... and N more files
• If file fetch failed or returned empty, print (files unavailable) instead

If a bucket is empty, omit its section entirely from the output.

If --ready-only is set, only the READY TO MERGE section is printed (plus the summary line).

Flags

Flag	Type	Default	Description
--repo	string	(all default repos)	Short name or full slug of a single repo to scan
--ready-only	bool	false	Print only READY TO MERGE PRs
--include-drafts	bool	false	Include draft PRs in all buckets

Error Handling

Situation	Action
gh pr list fails for one repo	Print [WARN] <repo>: gh fetch failed — skipping under that repo section; continue
All repos fail	Print error and exit
gh not authenticated	Print actionable message: gh auth login or check GH_TOKEN
No open PRs found in any repo	Print No open PRs found across scanned repos.

See Also

• review-all-prs skill — runs local-review on every open PR branch (heavy workflow)
• merge-sweep skill — merges all READY TO MERGE PRs
• auto-merge skill — merges a single PR with a Slack HIGH_RISK gate
• pr-watch skill — watches a single PR until CI passes or fails