Automation & Reports

Set the schedule.
Your bots do the rest.

Schedule complaint watches, dev reviews, red-team runs, guardrail audits, and automated reports. Attach Slack, email, webhooks, Discord or Teams. Every signal lands in the right channel, with the right recipients, at the right time — no on-call glue scripts.

Start for free See observability

Cron-simple, zero YAML

Five cadence presets (hourly → monthly) cover the 95% case. No crontab to memorize, no file to ship.

Five channels, one contract

Slack, email, webhook, Discord, Microsoft Teams — all reusable, all event-scoped, all togglable without rewiring tasks.

Signals already wired

Edge cases, red-team findings, and generated reports fire the right events by default — your channels pick up the right signals on day one.

The automation stack

Trigger → Task → Channel → Recipient

A single pipeline turns “I want a weekly ops report in Slack and the execs’ inbox” into something that actually runs forever.

The Automation Stack

trigger → task → channel

Triggers

cadence or signal

Tasks

7 scheduled types

Generators

reports + scans + red-team

Channels

slack · email · webhook · discord · teams

Recipients

filters + custom emails

Seven scheduled task types

ScheduledTaskType

Complaint Watch

complaint_watch

Watch real conversations for complaint signals — keyword triggers, risk score thresholds, and open alerts — so escalations never sit unattended.

Daily · keywords + risk ≥ threshold + includeOpenAlerts

Dev Review

dev_review

Roll up failing runs, pending insights, and open issues into a weekly packet engineers can act on — no more hunting through dashboards on Monday morning.

Weekly · failureThreshold + includePendingInsights

Trace Scan

trace_scan

Automatically scan recent traces for edge cases and regressions, surface them as candidates, and queue fixes in your evolution backlog.

Hourly · traceLimit + risk threshold

Red Team Run

red_team_run

Fire automated adversarial tests against a bot — jailbreaks, prompt injection, PII probing — and flag any weakness the moment it appears.

Weekly · strategies + testsPerStrategy

Guardrail Audit

guardrail_audit

Check that every enabled guardrail is still firing correctly and still relevant — detect drift between policy and behavior.

Weekly · all guardrails per bot

Report Generation

report_generation

Compile and distribute rich analytics reports on the cadence you choose, with the sections you choose, to the people you choose.

Monthly · executive summary + edge cases + red team

Custom Task

custom

Roll your own scheduled work — pass a customPayload, attach channels, and route results wherever your operators live.

Any cadence · freeform payload

Cadence — five presets, no cron editor

Hourly

every hour

0 * * * *

Daily

every day

0 9 * * *

Weekly

every week

0 9 * * 1

Bi-weekly

every two weeks

0 9 */14 * *

Monthly

every month

0 9 1 * *

Five notification channels, one contract

NotificationChannelType

Slack

slack

Post alerts and reports straight to any channel via webhook.

webhookUrl · optional headers

Send to specific recipients or org-scoped filters (all / admins / owners).

emails[] + recipientFilter

Webhook

webhook

Fan out to your own SIEM, ticketing tool, or on-call platform.

POST url + custom headers

Discord

discord

Ship into community or internal servers with one URL.

channel webhook

Microsoft Teams

teams

Deliver to Teams channels the enterprise way — no scripts.

incoming webhook

Default events on new channels

Channels subscribe to events. Every new channel starts listening for these three so you never miss the signals that matter most:

edge_case.created
A new regression or edge case was detected.
red_team.completed
A scheduled red team run produced findings.
report.generated
An automated report was compiled and is ready to distribute.

events[] is editable per channel

Reports

Eleven sections. Three cadences. Zero formatting work.

Pick the sections that matter, choose daily / weekly / monthly, and Rylvo compiles the right analytics window, renders the report, and delivers it to the resolved recipients.

Build a report in minutes

ReportSection[]

Executive summary

Plain-English headline of the window — what happened, what changed.

Edge cases

New regressions clustered by pattern, severity, and resolution status.

Guardrail performance

Trigger rates, blocks, and rewrites for every policy you shipped.

Red team findings

Adversarial runs — strategies used, hits, and remediation state.

Fix proposals

Agent Evolution candidates ready for review — with expected lift.

Improvements

Promoted rules and their measured delta vs the prior window.

Bot comparison

Side-by-side health across your active bots.

Complaints & escalations

Trigger reasons, sample conversations, and open alert state.

Developer follow-ups

Failing runs, pending insights, and items blocking the team.

Automation coverage

Which tasks / reports / channels are live, and what bots they cover.

Recommendations

Prioritized next-best actions tuned to your current risk profile.

recipient filter · custom emails

The right report. The right people. Every time.

Every report config has a role-scoped recipient filter and can merge in any email addresses you add directly. Attach email-type channels for team aliases. Everything resolves at send time so new hires and role changes are reflected automatically.

All members

Every member of the org receives the report on send.

Admins only

Restricted to members with the admin role.

Owners only

Highest-trust distribution — owners get the raw numbers first.

Custom list

Exact email addresses — works alongside email-type channels.

Failure loop

Detect → Cluster → Plan → Notify

Automation is more than cron. Rylvo classifies every failure, clusters patterns, maps them to evolution rules and guardrails, and pushes the fix plan to the right team — automatically.

Automatic failure detection → fix plan

failureDetection · failureResolution

Detect

failureDetection classifies every run against timeout, guardrail_block, parsing_error, loop_detected, and more.

Cluster

Similar failures collapse into FailurePatternDoc records with severity and counts.

Plan

failureResolution ties patterns to reviewer insights, evolution rules, and guardrails.

Notify

The right scheduled task forwards the fix plan to the right channel.

Where most teams start

one-click starters

Add a delivery channel

Without at least one channel, alerts have nowhere to land. Most teams start with Slack + a dedicated email.

Add Complaint Watch

Daily scan for escalation-worthy conversations — catches complaints before they catch you.

Add Dev Review

Weekly packet of failing runs and open insights, delivered to the engineering channel.

Create a Weekly Ops Report

Executive summary + edge cases + recommendations, auto-compiled and distributed every Monday.

Scope everything down to one bot, or keep it global

Per-bot scope

Attach any scheduled task or report to a specific bot — so complaint watches on your Support bot don’t fire against your Sales bot.

Global (all bots)

Or leave the bot field empty for org-wide coverage — ideal for executive reports, red-team sweeps, and guardrail audits that span your whole fleet.

Answers

What teams usually ask

What exactly can I schedule?

Seven task types: Trace Scan, Red Team Run, Report Generation, Guardrail Audit, Complaint Watch, Dev Review, and Custom. Each runs on your chosen cadence (hourly, daily, weekly, bi-weekly, monthly), can be scoped to a single bot or the whole org, and can attach any number of notification channels for delivery.

Which notification channels are supported?

Slack, Email, Webhook, Discord, and Microsoft Teams. Channels are reusable — create them once, attach them to any number of tasks or reports, and flip them on or off without re-wiring the tasks.

How do automated reports actually get built?

Pick a cadence (daily, weekly, monthly) and the sections you want — executive summary, edge cases, guardrail performance, red team, fix proposals, improvements, complaints & escalations, developer follow-ups, automation coverage, and recommendations. At send time Rylvo queries the right window (1 / 7 / 30 days), renders the sections, and distributes to resolved recipients.

Can I target different people for different reports?

Yes. Every report config has a recipient filter (all members / admins only / owners only / custom email list) plus any email-type channels you attach. Custom email addresses and filter-resolved org members are merged before delivery so the right people see the right report.

Is there automatic failure detection?

Yes. Runs flow through failureDetection, which classifies and clusters issues into failure patterns (timeout, guardrail_block, parsing_error, loop_detected, etc.). failureResolution links those patterns to reviewer insights, evolution rules, and guardrails — the pieces your scheduled tasks pick up and deliver to your team.

Does this talk to Mission Control and Agent Evolution?

Complaint Watch surfaces conversations worth opening in Mission Control. Dev Review pulls pending Agent Evolution insights into your weekly packet. Report sections include promoted rules, fix proposals, and open alerts — so your automation, supervision, and evolution loops share one signal graph.

Ready to set it and forget it?

Turn daily chores into standing orders.

Add a channel, attach a task, pick a cadence — your first automated report can be in the team’s inbox tomorrow morning.

Start for free See pricing