| name | cold-report |
| description | Generate a digest of cold outreach performance - deliverability, replies, booked calls, domain health. Use when the user asks for a report, digest, summary, weekly recap, performance check, or asks "how did my campaigns do". Reads live metrics from the FoxReach API. Designed to be called by cron (daily or weekly) and post results to Slack or email. Part of the cold.md suite. |
cold-report
Pulls live metrics from FoxReach and writes a digest. The last step in the cold.md loop - designed to run unattended on a schedule.
When to use
- User asks: "how did yesterday's sends do", "weekly report", "campaign digest", "what replied this week"
- Cron job fires this skill (daily 09:00 or weekly Mon morning)
- User says "reporting", "standup", "numbers"
Prerequisites
Output modes
Pick the mode based on user request. Default: terminal.
terminal - printed to stdout
Short, scannable digest for interactive use.
slack - posted to a Slack channel
For cron-driven posts. Takes SLACK_WEBHOOK_URL or posts via Buildberg Ops Bot if running inside the vault.
markdown - written to reports/YYYY-MM-DD.md
For archiving. Matches standup format.
Digest sections
Always include, in this order:
1. Window
The period covered. Default: last 24 hours. Weekly: last 7 days.
2. Send health
From GET /v1/campaigns/metrics?window={n}d:
- Emails sent (vs target send volume)
- Delivered / bounced / deferred counts
- Average send rate across inboxes
- Domain health flags - any domain with bounce rate > 3% or spam-rate > 0.1% gets a warning in red
3. Reply triage
From GET /v1/replies?window={n}d&group_by=bucket:
- Total replies
- Breakdown:
interested, not_now, ooo, bounced, unsubscribe, wrong_person, needs_review
- Flag if
needs_review is > 5 - nudge the user to run /cold triage.
4. Pipeline outcomes
From GET /v1/bookings?window={n}d:
- Booked calls count
- Conversion rate (bookings / interested replies)
- Top 3 campaigns by booking volume
5. Action items
Auto-derived from the metrics:
- If any domain has bounce rate > 3% → "Pause [domain] and investigate bounce cause"
- If
needs_review bucket > 5 → "Run /cold triage to sort these"
- If delivered count < 80% of target → "Sending capacity dropped - check inbox status"
- If weekly: "Replied but not booked: [N] leads - consider a breakup in cold-draft"
Template
COLD.MD REPORT - {window}
{date-range}
---
SEND HEALTH
Sent: {sent} / {target} ({pct}%)
Delivered: {delivered} ({pct}%)
Bounced: {bounced} ({pct}%) {warn if >3%}
Avg/domain: {rate}/day
REPLY TRIAGE
Interested: {n}
Not now: {n}
OOO: {n}
Bounced: {n}
Review: {n} {warn if >5}
PIPELINE
Booked: {n}
Conv rate: {n}%
Top campaign: {name} ({n} bookings)
ACTIONS
- [action 1]
- [action 2]
---
Full dashboard: https://foxreach.io/app
Colors in terminal mode (fallback gracefully on non-tty or NO_COLOR):
- Rust-orange (
\033[38;2;194;65;12m) for headers and numbers
- Red (
\033[31m) for warnings
- Green (
\033[32m) for "healthy" indicators
- Dim grey for labels
Scheduling this skill
Via FoxReach (recommended)
FoxReach dashboard > Notifications > Daily digest. Posts to Slack/email. Zero setup on the user's side.
Via Claude Code schedule skill
/schedule "0 9 * * *" "cold report daily"
/schedule "0 9 * * 1" "cold report weekly"
Requires user's laptop to be on + Claude Code running. Power-user only.
Via a local cron
0 9 * * * cd /path/to/project && FOXREACH_API_KEY=... claude "cold report daily --slack" 2>&1 | logger -t cold-report
Guardrails
- Never fabricate metrics. Only report what the API returned. If an endpoint is unreachable, say so explicitly.
- Round thoughtfully. Open rates to whole percentages. Bounce rates to one decimal.
- Respect time zones. Default is user's local time zone from
$TZ; surface the zone in the header.
Experiment outcomes (autoresearch)
If .cold/decisions.md exists, append an "Experiments" section to the digest covering all decisions since the last report (track via a timestamp in .cold/last-report.json).
For each concluded experiment, include:
### Tier <N> / <variable>
- Experiment: <id>
- Outcome: winner = <arm> | inconclusive
- Numbers: <arm_a> 5.0% vs <arm_b> 8.0% interested-reply rate
- Confidence: 95% CI [+0.2, +5.8] (p=0.04)
- Status: applied via diff | pending review | rejected
- Belief now locked: <one sentence>
If a diff is pending review, surface it prominently:
⚠ ACTION NEEDED: cold.md diff awaiting your review since <date>
.cold/proposed-diff.patch — accept with `git apply` or reject with `rm`
If an experiment is currently in flight, include a "Currently testing" line with days remaining.
After printing, update .cold/last-report.json:
{ "lastReportAt": "2026-04-29T...", "type": "weekly" }
References
