ThroughloopThroughloop
Throughloop workflows

Pre-built loops the agent runs for you.

Each workflow is a closed loop: a precise trigger, scoped reasoning, bounded actions, and a guardrail. Add yours next.

PR Review Recovery
When a PR sits in review longer than your team's SLA, route a backup reviewer and notify the owner — before standup catches it.
active
Trigger
PR waiting > 24h
Reasoning
Reviewer is overloaded (>3 pending PRs). Find an available engineer with relevant context.
Actions
  • Assign backup reviewer
  • Notify #engineering with context
  • Update Linear ticket status
Guardrails
  • Require approval before external messages
  • Skip if PR author is OOO
  • Never reassign during a freeze window
14 runs · 93% success
last run · 2m ago
Sprint Drift Detection
Continuously evaluates sprint commitment vs. burn-down. Flags drift early enough to actually do something about it.
active
Trigger
Burn-down > 1.4σ from plan
Reasoning
Compare ticket completion velocity, review queue depth, and remaining scope. Surface the 1–2 tickets pulling the sprint off-track.
Actions
  • Annotate sprint with drift summary
  • Suggest scope cuts to PM
  • Schedule a 10-min sync if drift > 2σ
Guardrails
  • Never auto-remove tickets
  • Always show reasoning before suggesting cuts
6 runs · 88% success
last run · 26m ago
Blocked Engineer Escalation
When an engineer is blocked on review, dependency, or env, escalate to the right unblocker — not the whole channel.
active
Trigger
Engineer blocked > 4h
Reasoning
Identify the precise dependency. If it's a review, route to backup reviewer. If it's a dep, ping the owning team lead privately.
Actions
  • DM the unblocker with context
  • Mirror to #engineering only if unresolved in 1h
  • Update standup notes
Guardrails
  • Private DM before public channel
  • Never page on-call outside business hours unless severity = high
21 runs · 95% success
last run · 11m ago
Incident Follow-up Generator
After an incident closes, auto-draft the postmortem skeleton, owner-assign action items, and schedule the review meeting.
draft
Trigger
Incident severity ≥ 2 resolved
Reasoning
Pull commits, alerts, and Slack thread. Draft a Five Whys, propose action items, assign owners.
Actions
  • Draft postmortem doc
  • Create follow-up tickets
  • Schedule review on calendar
Guardrails
  • Human approval before publishing postmortem
  • Owner can decline assignment in 1 click
3 runs · 100% success
last run · 2d ago
Coming soon
Workflows we’re building next, ranked by user demand.
  • On-call rotation balancer
    Detects skewed pages → suggests rotation tweaks.
  • Stale-PR auto-rebase
    Rebases drifting PRs on main, runs CI, hands back.
  • Spec-to-tickets generator
    Reads PRDs, drafts a ticket tree, tags owners.
  • Deploy risk scorer
    Predicts incident-prone PRs from diff + history.