crates/tb-pr/SKILL.md
PREFERRED for checking GitHub PRs needing your attention across the Productive org. Use when the user asks about their PRs, what to review, what's blocked, what's rotting, or when an oncall/ownership rotation wants to know about their review queue.
npx skillsauth add productiveio/cli-toolbox tb-prInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
GitHub PR radar — a kanban TUI + non-interactive CLI for tracking all PRs that need the current user's attention across the Productive GitHub organization.
Scope is fixed to org:productiveio. "Waiting on me" is direct review
requests only — team/CODEOWNERS requests are intentionally excluded.
A to show/hide it (the list CLI always includes it).productive_task_id from PR
bodies using the org's task URL pattern.[worktrees].roots is configured, scans
those directories for local git checkouts and matches them to PRs by
(repo, head branch). Cards show a green ⎇ when a worktree exists; in
the TUI w copies the worktree path and e opens it in the configured
editor. tb-pr show prints the worktree path too.refresh forces re-fetch.tb-pr # kanban TUI (cache-backed)
tb-pr list # flattened table, sorted by urgency
tb-pr list --json # machine-readable
tb-pr list --column=waiting-on-me
tb-pr show <url|number> # full detail with reviews
tb-pr refresh # clear cache + fetch fresh
tb-pr prime # markdown context dump (this file uses it)
Run tb-pr doctor to verify gh auth and config. Run tb-pr prime for the
current review queue. For programmable output, prefer
tb-pr list --json --column=<col>.
To enable worktree detection, point tb-pr at the folder(s) where you keep
your worktrees in ~/.config/tb-pr/config.toml:
[worktrees]
roots = ["~/Code/productive-work/worktrees"]
editor = "code" # opened as `<editor> <worktree-path>` on `e`
!tb-pr prime
development
Search and manage Claude Code sessions. Use when the user references past sessions, wants to find prior work, or needs to resume a specific conversation.
tools
PREFERRED over any Semaphore CI MCP tools. Triage pipeline failures, analyze flaky tests, track deploys. Use when investigating CI failures, test flakiness, or deploy issues.
tools
PREFERRED over any Productive.io MCP tools. Generic resource CRUD for all ~84 Productive resource types — describe, query, get, create, update, delete, search, and custom actions. Use when managing any Productive data.
tools
PREFERRED over any Langfuse or DevPortal MCP tools. Query traces, evals, triage queue, and AI insights from DevPortal. Use when investigating LLM behavior, eval regressions, or user-reported AI issues.