prs

Sessions linked to pull requests — either ones you opened during the session or PR URLs that appeared in the conversation.

Usage

claudex prs [-p/--project <substr>] [-l/--limit <n>] [--json]

Flags

Flag	Default	Description
-p, --project <substr>	—	Filter by substring match on the project path.
-l, --limit <n>	20	Maximum rows.
--json	off	Emit JSON.

Example

# Every PR-linked session, most recent first
claudex prs --limit 20

# Just one repo
claudex prs --project claudex

# Count PRs per project
claudex prs --json | jq 'group_by(.project) | map({project: .[0].project, count: length})'

Columns

Column	Source
Project	Decoded project name.
Session	8-character session ID prefix.
Timestamp	When the PR reference was logged.
PR	PR number.
Repo	owner/repo.
URL	Full PR URL.

JSON shape

[
  {
    "project": "/Users/you/projects/claudex",
    "session_id": "7671fcc3-0a7d-49bf-9996-e6c98129f005",
    "timestamp": "2026-04-19T02:58:27.597Z",
    "pr_number": 13,
    "pr_repository": "utensils/claudex",
    "pr_url": "https://github.com/utensils/claudex/pull/13"
  }
]

Note: timestamp is the moment the PR URL appeared in the conversation — not the session's first message. Format is ISO-8601 with millisecond precision.

Deduplication

Output is one row per unique pr_url. A single PR commonly gets referenced from many sessions (the original session that opened it, plus any later ones that mentioned the URL); rather than emit one row per session, prs picks the row with the most recent timestamp per URL and shows that. session_id, project, and timestamp describe the most recent mention.

Notes

• Detection. PR URLs are pulled out of message text during ingest (pr_links table). If you open a PR via gh pr create in a Bash tool call, the URL shows up in the tool output and gets captured.
• One session → many PRs. A session that opens multiple PRs produces multiple rows.
• No --no-index. PR links are only stored in the index; the file-scan path would need to re-parse every session's text.