Tag compliance
Mandatory, untagged, prohibited, tag-limit and allowed-value checks across every taggable resource type, so spend can be allocated.
spendlens tags --summary
FinOps for AWS, from your terminal
spendlens finds untagged resources, oversized instances, idle waste and unexpected cost spikes, puts a price on each one, and answers your questions in plain English. Read-only, by design.
$ spendlens waste --summary --by service service total alarm compliance est. saving/mo ebs 48 7 85.4% $507.75 rds 6 1 83.3% $49.64 elb 6 2 66.7% $32.86 14 alarm(s) on 14 resources · est. saving USD 595.10/month $ spendlens ask "Why did our bill go up last month?" › history_query · 2 queries · read-only Spend rose +$959 (+1.8%) to $55.4K in October. EC2 grew +$3,698 and Amazon Bedrock is new (+$1,660), partly offset by Redshift being decommissioned (−$3,710). Source: cost_diff run #7, 2026-11-09.
Sample output from the bundled demo data (--source demo).
Every check runs against your real AWS estate through Steampipe, with your company's own standards
from a single finops.yaml.
Mandatory, untagged, prohibited, tag-limit and allowed-value checks across every taggable resource type, so spend can be allocated.
spendlens tags --summary
Oversized instances, gp2 volumes, previous generations, Multi-AZ in non-prod, versions in paid Extended Support. Each one priced.
spendlens rightsize --min-saving 50
Unattached volumes, orphaned snapshots, idle load balancers and NAT gateways, unused IPs and secrets: billed, but doing nothing.
spendlens waste
What changed between any two months or quarters, by service, account, region or tag. New and removed services always included.
spendlens cost diff -g tag:Team
Daily spikes against a robust baseline, and a month-end projection checked against your budget and every AWS Budget.
spendlens anomalies
AWS Cost Optimization Hub, Savings Plan coverage and spendlens's own findings, merged without double counting.
spendlens savings summary
Claude, OpenAI, Azure or any OpenAI-compatible model answers from your data, and can never change anything.
spendlens ask "why did EC2 go up?"
A self-contained HTML dashboard, and table, CSV, JSON or HTML output from every command.
spendlens dashboard -o report.html
One-command runs in GitLab or GitHub on a read-only OIDC role, with Slack and Teams summaries of what changed.
spendlens init --ci gitlab
Every threshold is a parameter in one finops.yaml, overridable per run. Plugins add controls, commands and AI tools.
spendlens params
From the weekly platform review to the month-end forecast, spendlens fits how engineering and finance already work.
Start the meeting with what changed and what's flagged, then drill into the services behind it.
spendlens cost diff --show flagged
spendlens waste --summary --by service
Showback by team and environment, with the untagged share called out so it can be fixed at the source.
spendlens cost showback -t Team
spendlens tags -k mandatory
Projected month end against the budget, and a savings total you can defend line by line.
spendlens forecast --fail-on-risk
spendlens savings summary
Fail a pipeline on tag-policy violations or big cost jumps, and post the result to the team channel.
spendlens tags --fail-on-alarm
spendlens notify --slack --only-if-changes
Find the anomaly, then ask the assistant to explain it from your own data, with the queries it used.
spendlens anomalies
spendlens ask "what drove yesterday's spike?" --show-tools
Allowed instance families, Graviton adoption, gp3 everywhere, Multi-AZ only in prod, all from one policy file.
spendlens rightsize -S ec2 -S rds -S ebs
Uses whatever AWS role you already have assumed. No agents, no SaaS, no stored keys.
spendlens plus Steampipe and Powerpipe.
pip install "spendlens[ai] @ git+https://github.com/devops-monk/spendlens.git"
brew install turbot/tap/steampipe turbot/tap/powerpipeMandatory tags, allowed instance types, budgets.
spendlens init
spendlens doctorTry it on sample data first, then point it at your account.
spendlens report --source demo
spendlens init --ci gitlabspendlens never creates, changes, tags or deletes anything. Savings are proposals for a human to review. In CI it assumes a read-only role through OIDC, so no AWS keys are stored. Its AI assistant can only run SQL that passes a parser and then executes on connections that cannot write.