Skip to content

Fawkes DORA Metrics — Rework Rate

AGENTS.md reference: Section 9 — Fawkes-Specific Principles
Rework rate is tracked here and checked weekly via scripts/weekly-metrics.sh.

1. Rework Rate Definition

Rework rate measures the percentage of pull requests that required follow-up fixes within seven days of merging (bug fixes, reverts, or hotfixes traceable to a recently merged PR).

Threshold Status Action
< 10 % 🟢 GREEN — Healthy No action required. Continue current practices.
10 – 20 % 🟡 YELLOW — Watch Review recent PRs for patterns. Schedule a retro item.
> 20 % 🔴 RED — Stop features Halt new feature work. Conduct a root-cause analysis.

These thresholds align with DORA's definition of change failure rate and are applied to the Fawkes mono-repo across all layers (services, infra, platform, scripts, docs).

2. How to Read the DevLake Rework Dashboard

  1. Open Grafana at http://devlake-grafana.127.0.0.1.nip.io (local) or the environment URL configured in your GRAFANA_URL environment variable.
  2. Navigate to Dashboards → DORA Metrics → Rework Rate.
  3. Set the time range to Last 7 days (use the date picker in the top-right corner).
  4. Key panels:
  5. Rework Rate (%) — headline percentage for the selected period.
  6. Rework PRs — list of individual PRs counted as rework.
  7. Trend — 12-week rolling chart; look for sustained upward movement.
  8. Hover over any bar in the Rework PRs panel to see the PR title, author, and the original PR it is fixing.

3. Current Baseline

Status as of last update: TBD — baseline not yet established.

Week Rework Rate Status Notes
TBD TBD % TBD Awaiting first DevLake data collection cycle

This table is updated automatically by scripts/weekly-metrics.sh during the weekly metrics review run.

4. Weekly Review Process

The weekly metrics review ensures that rework rate stays within the healthy threshold and that any degradation is caught early.

When

Every Monday at 09:00 UTC (or the first working day of the week).

Who

  • Platform Engineering lead (owns the review)
  • One rotating team member (shadow reviewer)

Steps

# 1. Run the metrics script — it queries DevLake and updates this file
./scripts/weekly-metrics.sh

# 2. Check the terminal output for the traffic-light status
#    GREEN  → no action
#    YELLOW → add retro item to this week's board
#    RED    → create a P1 issue, pause new features

# 3. Commit the updated baseline table (section 3 above) with message:
#    chore(metrics): weekly rework rate update YYYY-MM-DD

Escalation

Status Escalation path
🟢 GREEN No escalation
🟡 YELLOW Post summary in #platform-engineering Mattermost channel
🔴 RED Page on-call lead, open a P1 issue, freeze feature merges via branch protection rule

The following DORA metrics are tracked alongside rework rate. See the DevLake DORA dashboard for full details.

Metric Target (Elite) Source
Deployment Frequency ≥ 1/day DevLake / ArgoCD events
Lead Time for Changes < 1 hour DevLake / GitHub PRs
Change Failure Rate < 5 % DevLake / incident records
MTTR < 1 hour DevLake / PagerDuty
Rework Rate < 10 % DevLake / GitHub PRs

6. See Also

  • scripts/weekly-metrics.sh — automated weekly data collection
  • docs/runbooks/ — incident runbooks
  • docs/AGENTS.md Section 9 — platform principles
  • DevLake documentation: https://devlake.apache.org/docs

AI-Readiness Metrics

Last updated: 2026-03-13 by scripts/check-ai-readiness.sh Threshold: >= 80% GREEN | 50-79% YELLOW | < 50% RED

Service Type-hints Docstrings Unit Tests BDD
ai-code-review 3/10 (30%) 10/10 (100%) N Y
analytics-dashboard 9/27 (33%) 27/27 (100%) N Y
anomaly-detection 3/12 (25%) 12/12 (100%) N Y
devex-survey-automation 1/21 (4%) 21/21 (100%) N N
discovery-metrics 0/24 (0%) 24/24 (100%) N Y
experimentation 13/34 (38%) 34/34 (100%) N Y
feedback-bot 5/9 (55%) 9/9 (100%) N Y
feedback 21/43 (48%) 40/43 (93%) N Y
friction-bot 2/8 (25%) 8/8 (100%) N N
insights 2/23 (8%) 23/23 (100%) N Y
mcp-k8s-server 0/3 (0%) 0/3 (0%) N N
nps 2/11 (18%) 11/11 (100%) N N
rag 0/8 (0%) 8/8 (100%) Y Y
smart-alerting 13/30 (43%) 30/30 (100%) N Y
space-metrics 8/21 (38%) 21/21 (100%) N Y
vsm 3/15 (20%) 15/15 (100%) N N