SECOPS
Auto-triage and label sensitive GitLab MRs with a security agent
An agent reads each newly opened merge request, judges its security sensitivity from the diff and description, applies the right security labels.
How it runs
The automated pipeline, trigger to output.
- TriggerGitLab merge request openedGitLab
- ActionFetch MR diff, title, and descriptionGitLab
- LogicAgent classifies sensitivity tierOpenAI
- LogicBranch sensitive vs low-risk
- ActionApply labels and assign reviewer poolGitLab
- OutputComment classification rationale on MRGitLab
What it does
Replaces manual security triage with an agent that reads the full context of every new merge request, the diff, title, and description, and decides its sensitivity. It applies the appropriate `security::*` labels (crypto, authn, pii, infra, or none), writes a short rationale comment, and assigns the matching reviewer group when the change is sensitive.
When to use it
Use it when MRs arrive unlabeled and humans inconsistently flag what actually needs security review, so risky changes slip through as ordinary work. The agent gives every MR a consistent, explained classification on arrival.
How it works
- 1A GitLab webhook fires when a merge request is opened.
- 2The agent fetches the diff, title, and description for the MR.
- 3It reasons about which sensitive surfaces the change touches and assigns a sensitivity tier with a rationale.
- 4A branch separates sensitive MRs from low-risk ones.
- 5For sensitive MRs it applies labels, comments its reasoning, and assigns the right reviewer pool in GitLab; low-risk MRs get a `security::reviewed-auto` label and a closing note.
Set it up
What you configure once, before turning it on.
- 1Connect GitLabRepos, MRs, pipelines, registry.
- 2Connect OpenAIModels, embeddings, files.
- 3Set each agent's modelWe leave models unset so you pick the tier — fast + cheap, or top-quality.
- 4Tune it to your dataEdit the prompts, filters, and field mappings so it matches how your team works.
- 5Test, then turn it onRun once against a sample, confirm the output, then enable the trigger.
More SecOps workflows
Compile a weekly WAF tuning review with trends to Confluence
Every week an agent rolls up Cloudflare WAF block clusters by rule and ASN, compares them to prior weeks for trend direction.
Sensitive Dropbox Link Owner Remediation Loop
When a newly created Dropbox shared link points to a sensitive file, this workflow DMs the file owner, gives them a deadline to justify or revoke it.
Post-Revocation Verification and Audit Logging
After a key is revoked, it confirms the old credential actually fails, verifies the replacement works.
Exposed-Secret Incident Triage and Remediation Agent
An agent-driven workflow that investigates a reported leaked secret end to end, decides revoke-versus-escalate, executes the rotation.
Page on-call when a WAF rule mass-blocks legitimate traffic
On demand or every few minutes, it detects a single Cloudflare WAF rule suddenly blocking a broad spread of ASNs and paths (a likely false-positive storm).
PII Content Scan on New Dropbox External Share
When a file gets an external Dropbox link, it reads the file content, uses an AI classifier to detect PII or secrets.
Run it inside a business
This workflow drops into a full company template. Import the org, and this is one of the playbooks its agents run.
Run this workflow in your colony.
14-day trial. No DevOps. No Sales call. Provisioned in under a minute.