rootly-morning-brief

OpenClaw skill for a short daily Rootly briefing:
- active incidents
- incidents resolved in the last 24 hours
- who is on-call now
- overdue action items

The script prints the digest to stdout. OpenClaw cron --announce is what delivers that output to Slack.

Phone Screenshot

This repo includes a real 8:00 AM phone delivery screenshot from the RootlyClaw app:
- Output-screenshot.jpg

Prerequisites

• OpenClaw is installed and running.
• Python 3 is available on the host or in the sandbox image.
• For live Rootly data, you have a Rootly API key.
• For Slack delivery, Slack is already connected in OpenClaw.

If you run OpenClaw in a sandboxed container, make sure that container includes python3. OpenClaw sandboxing provides isolation, not a built-in Python runtime for skills.

Install

ClawHub:
- Install the ClawHub CLI if you do not have it yet:

npm i -g clawhub

or

pnpm add -g clawhub

• Install the skill:

clawhub install rootly-morning-brief

• Start a new OpenClaw session so the skill is picked up.

By default, clawhub installs into ./skills under your current working directory. If an OpenClaw workspace is configured, it falls back to that workspace and OpenClaw loads the skill from <workspace>/skills.

Manual install:
- Copy this folder to ~/.openclaw/workspace/skills/rootly-morning-brief
- For a named profile, use that profile's workspace instead

Configure the Rootly API Key

You only need this for live Rootly data. Mock-data runs do not need an API key.

Recommended for OpenClaw:

mkdir -p "${OPENCLAW_STATE_DIR:-$HOME/.openclaw}/secrets"
printf '%s\n' 'YOUR_ROOTLY_API_KEY' > "${OPENCLAW_STATE_DIR:-$HOME/.openclaw}/secrets/rootly_api_key"
chmod 600 "${OPENCLAW_STATE_DIR:-$HOME/.openclaw}/secrets/rootly_api_key"

OpenClaw cron jobs do not inherit your shell env, so the secret file is the reliable path.

For temporary manual testing, this also works:

export ROOTLY_API_KEY="YOUR_ROOTLY_API_KEY"

Optional:

export ROOTLY_TIMEZONE="America/Toronto"
export ROOTLY_INCLUDE_PRIVATE="false"

Step 1. Test Local Output with Real Data

This is the main path for the real daily use case.
Requires the Rootly API key from the section above. If you do not have one yet, skip to Step 2 (mock data).

python3 scripts/rootly_morning_brief.py

Step 2. Test Local Output with Mock Data

Use this when you want to verify formatting without hitting the API.

python3 scripts/rootly_morning_brief.py --mock-data-dir ./mock-data

./mock-data is relative to the skill directory.

Step 3. Test Slack Delivery Once

The script does not send to Slack by itself. OpenClaw cron --announce captures stdout and sends the briefing.

Start with the simplest path from the current or last chat context:

openclaw cron add \
  --name "Rootly morning brief test" \
  --at "+1m" \
  --delete-after-run \
  --session isolated \
  --light-context \
  --message "Use rootly-morning-brief. Run scripts/rootly_morning_brief.py and print the full digest." \
  --announce

If you want to pin a specific Slack channel instead:

openclaw cron add \
  --name "Rootly morning brief test" \
  --at "+1m" \
  --delete-after-run \
  --session isolated \
  --light-context \
  --message "Use rootly-morning-brief. Run scripts/rootly_morning_brief.py and print the full digest." \
  --announce \
  --channel slack \
  --to channel:C0123456789

If you want the one-time Slack test to use mock data, make the instruction explicit:

openclaw cron add \
  --name "Rootly morning brief test (mock)" \
  --at "+1m" \
  --delete-after-run \
  --session isolated \
  --light-context \
  --message 'Use the `rootly-morning-brief` skill. Run it with mock data from `./mock-data` and print the digest.' \
  --announce

You can wait a minute or force due jobs immediately:

openclaw cron run --due

Step 4. Create the Daily Cron Job

After the one-time Slack test works, register the real daily job.

The requirements example uses America/New_York. This skill defaults to America/Toronto on purpose because that is the timezone the script and sample data were built around. Change it if your team wants a different local morning window.

Current or last chat context:

openclaw cron add \
  --name "Rootly morning brief" \
  --cron "0 8 * * *" \
  --tz "America/Toronto" \
  --session isolated \
  --light-context \
  --message "Use rootly-morning-brief. Run scripts/rootly_morning_brief.py and print the full digest." \
  --announce

Explicit Slack channel:

openclaw cron add \
  --name "Rootly morning brief" \
  --cron "0 8 * * *" \
  --tz "America/Toronto" \
  --session isolated \
  --light-context \
  --message "Use rootly-morning-brief. Run scripts/rootly_morning_brief.py and print the full digest." \
  --announce \
  --channel slack \
  --to channel:C0123456789

Verify

List jobs:

openclaw cron list

Inspect runs for a job:

openclaw cron runs --id <job-id>

Run due jobs right now:

openclaw cron run --due

Sample Output

*Rootly Morning Brief* — Sun Mar 15
At a glance: 2 active (1 SEV0/SEV1) · 1 resolved in 24h · 2 on-call now · 1 overdue

*Active now*
• 🚨 <https://root.ly/gsif-3|Global sign-in failures after OIDC key rotation> — [SEV0] · [OPEN] · started Sun 6:42 AM
• 🟧 <https://root.ly/clsf-2|Checkout latency spike during us-east database failover> — [SEV2] · [OPEN] · started Sun 5:25 AM

*On-call now*
• Nicole Bu — L1 primary
• Jordan Patel — L2 secondary

*Overdue actions*
• ⚠️ <https://root.ly/gtb2es|Rotate CI deploy tokens and verify revocation in every production region.> — [P1] · due Sat 7:30 AM · Nicole Bu · SEC-742

*Resolved (24h)*
• <https://root.ly/ubaa-1|Unauthorized bastion access attempt blocked> — resolved Sun 12:02 AM

Judgment Calls

• Phone-first structure (simplified on purpose): one top summary line + four fixed sections.
  Why: on-call readers need fast scanability; alerting guidance emphasizes concise, actionable signal over dense context.
• Action-first ordering: active incidents, who is on-call now, overdue actions, then resolved(24h).
  Why: this mirrors immediate operational questions before historical context.
• Tight line grammar: title — status/severity/time with one line per item.
  Why: reduces cognitive load and supports sub-30-second mobile reading.
• Severity and urgency visuals (non-flashy): SEV-based badges (🚨, 🟥, 🟧) and priority chips ([P1]) on action items.
  Why: faster triage scanning under pressure without adding long text.
• Always render all sections: explicit No ... lines instead of hiding empty groups.
  Why: avoids ambiguity between “none” and “failed to load.”
• Default cap = 3 items per section: show top items, then +N more.
  Why: prevents long digests from becoming pager-noise walls.
• Only actionable metadata in the body: assignee + due time for overdue actions; ticket key only when obvious.
  Why: keep decision-relevant fields, skip verbose payload details.
• Delivery abstraction: script prints to stdout only; OpenClaw cron --announce handles Slack routing.
  Why: cleaner separation of concerns and easier runtime debugging.
• Explicit simplification vs full API coverage: only /v1/incidents, /v1/oncalls, /v1/action_items are used.
  Why: exactly matches required morning brief questions without overbuilding.

References used for formatting decisions:
- Google SRE guidance on actionable alerting and reducing noisy/non-actionable signal: https://sre.google/sre-book/practical-alerting/
- Atlassian incident communication tips (communicate early/often/precisely): https://support.atlassian.com/statuspage/docs/incident-communication-tips/
- Atlassian incident response handbook (short customer updates, explicit impact/next steps): https://www.atlassian.com/incident-management/handbook/incident-response

What This Skill Does Not Set Up

• It does not connect OpenClaw to Slack for you.
• It does not manage its own Slack delivery. OpenClaw cron --announce handles that.
• It does not discover the correct Slack destination automatically.
• It does not make cron inherit shell env vars.

This skill is responsible for fetching Rootly data and formatting the briefing. OpenClaw is responsible for chat delivery and cron routing.