PR Blocker Summarizer: Turn PR Noise into a Standup Digest

Where this fits

You're a tech lead with 15 open PRs across three repos. Your morning standup needs a quick read on which PRs are blocked, which are ready, and what's stale. Manually scanning GitHub tabs is slow and error-prone—you want an agent to parse a PR export and hand you a digest.

Why agents benefit

• Parses a JSON export of PRs and classifies each as blocked, ready, or stale based on checks, review status, and age.
• Emits a one-line summary (e.g., "3 blocked, 2 ready, 5 stale") that an agent can post directly to Slack or Teams.
• Handles missing fields conservatively, so partial exports don't crash the pipeline.
• Activates only on specific intents like "summarize open PRs" or "what's blocking my PRs"—no false triggers on general GitHub questions.
• Produces a deterministic output that can be fed into downstream standup automation or reporting tools.

Practical scenarios

Morning standup prep

Feed the skill a JSON export of all open PRs. It returns a blockers-first digest: which PRs have failing checks, which have changes requested, and which have been idle for over a week. Your agent can then post that digest to the team channel before the meeting starts.

Review backlog triage

When you have dozens of PRs awaiting review, the skill highlights those with pending reviews and no activity. You can prioritize the ones with the longest wait time or those with approved reviews but failing checks—so you know exactly where to apply pressure.

Stale PR cleanup

The skill flags PRs that are old and untouched. Use that list to close or rebase stale branches, or to nudge authors. This keeps your repository from accumulating zombie PRs that slow down CI and confuse contributors.

Add it to your agent workflow

Install the skill with:

npx skills add FrancyJGLisboa/agent-skill-creator --skill pr-blocker-summarizer

Then provide a JSON array of PRs. Example prompt:

Summarize these PRs for standup:
[{"title":"Add auth middleware","state":"open","checks":"failing","review":"pending","age_days":3}, ...]

The skill returns a compact digest with counts and a list of blocked items, ready for your agent to relay.

For more context, see the skill's page at https://www.openagentskill.com/skills/francyjglisboa-pr-blocker-summarizer.

Compare before adopting

Before wiring this in, compare it against your own hand-rolled triage script. Check the accuracy of its classification logic (does it treat missing checks as passing or failing?), how recently it was reviewed (last_reviewed: 2026-07-20), and whether it fits your PR export format. Also consider alternatives like a generic summarizer that might not have the blockers-first focus.

Why it is worth tracking

This skill is a small but well-crafted example from the agent-skill-creator repository (2,332 stars). It demonstrates a clean activation pattern, conservative handling of missing data, and a focused output—all qualities worth emulating. Track it to see how the author iterates on the bundling and eval pipeline, and evaluate it again if your PR triage workflow grows more complex.