$~:whoami> matheo, it & cybersecurity

I build agentic products.
Secure by design. Lean by default. Built to deliver.

Reformed tech bro, world explorer, mentor. Pushing to prod since Win95. Today I orchestrate fleets of AI agents that ingest, verify, write, and ship, with a human at every gate that matters.

Meet the agentsSelected work

Press ~ for a terminal.

$~:ls work/

Things I build and run.

Cyber threat intelligence

SitRep

A daily security brief for people drowning in security news. Curated sources go in; an agent workforce enriches, verifies, and writes the brief that comes out.

  • Multi-agent pipeline
  • Editorial quality gates
  • Free core service

Giving back

Cyber Seniors

A pro-bono initiative that helps older adults recognize scams, protect themselves online, and know where to turn when cybercrime hits.

  • Workshops
  • Free resources
  • Community

$~:ls agents/

My team is mostly agents.

Two products, one engineering loop. Every agent has a single job, writers never grade their own work, and numbers come from data, not models. I frame the work and own the ship gate.

Matheo

Orchestrator · frames the work, owns the ship gate

SitRep

An autonomous newsroom that turns the security firehose into a daily brief.

Agent roles
12
Pipeline stages
6
Scheduled jobs
19
  • Scraper & Scheduler FleetContinuous

    Keeps every source pipeline alive and re-fetches stale sources.

    • Feed IntakeEvery 4h

      Collects, normalizes, and stages articles from monitored sources.

  • Citation ValidatorPer article

    Keeps primary sources only: advisories, research, wire services.

    • Citation ExtractorPer article

      Pulls referenced URLs, bylines, and publication metadata.

  • Editorial ReviewPer article

    Applies quality checks so the feed stays actionable.

    • SummarizerPer article

      Writes an impact-led summary, tags, and an editorial priority score.

    • Buzz CreatorDaily

      Synthesizes ranked articles into the day’s narrative.

    • Rollup PublisherWeekly+

      Assembles weekly, monthly, and quarterly narratives.

  • Email EditorPer brief

    Quality gates: length, verified-claim rate, dedup. Escalates on failure.

    • Email CuratorPer brief

      Picks the lead story and drafts the editorial thesis.

    • Email WriterPer brief

      Drafts every section of the brief.

    • IOC SanitizerPer brief

      Defangs every indicator of compromise before it ships.

Live workforce dashboard ↗

Navigator

A relocation research tool where AI answers are grounded, bounded, and checked.

AI pipeline roles
6
Scheduled workers
41
Freeform agent loops
0
  • Intent RouterPer question

    Classifies the question and routes it to the right tools.

    • PlannerPer question

      Plans which data lookups the answer needs.

    • ComposerPer question

      Writes the answer from tool outputs only.

    • Numeric FinalizerPer question

      Every number comes from a calculator or the database, never the model.

  • Brief VerifierPer narrative

    Checks every claim against evidence and edits what it can’t support.

    • City NarratorMonthly

      Writes evidence-only city profiles.

  • Ops crewDaily → monthly

    Job-health watchdog, AI quality sweep, retention and deletion.

    • Ingestion crewDaily → yearly

      27 workers keep official and public datasets current.

    • Enrichment crewWeekly → monthly

      7 workers add context to ingested records.

Engineering loop

How both products get built: agents ship, other agents grade, I own two gates.

Human gates
2
Maker = checker
Never
  • Next ItemPer cycle

    Finds the highest-leverage work and writes verifiable success criteria.

    • Loop StateEvery stage

      Keeps on-disk memory truthful so no context is lost between sessions.

  • ReviewerPer change

    Grades goal fit, simplicity, scope, and safety. Never the author.

    • MakerPer change

      Implements the item: simple, surgical, test-first.

    • Security AuditorPer change

      Audits input handling, auth, data leakage, and rate limits.

    • Verify ItemPer change

      Evidence before claims: runs the app and captures the result.

  • Ship ItemPer change

    Opens the PR with evidence and updates the backlog.

    • CI guardsPer PR

      Tests, dependency review, and security scans on every PR.

  • AI
  • Automation
  • Deterministic
  • Checker
  • Reports to the checker above

$~:ls arcade/

Off the clock.

Four hand-coded games in a pixel arcade, and a terminal with a few secrets. Try whoareyou.

Open the arcade