Skip to content
@workloftai

Workloft

How much can one person build with a fleet of AI agents? Experiments, research and builds, shared in the open.
README.md

Workloft

How much can one person build with a fleet of AI agents?

Workloft is a one-person AI shop. I run a fleet of agents on a sovereign VPS (my own server, nothing rented from a hyperscaler) and point them at hard problems: agent infrastructure, model routing, verification, and tooling for regulated UK buyers (councils, FCA-regulated firms, healthcare, education).

This org is where the open work lives: the runtimes, the guardrails, the eval harnesses. Experiments, research and builds, shared as I go.

What's here

  • civiclaw : open-source, audit-native agent runtime for the UK public sector (DSAR, FOI, EIR).
  • loop-pilot : a todo system the agent cannot cheat. A Claude Code Stop hook plus a TTL escalator.
  • auto-rubrics : auto-generate evaluation rubrics from agent audit-log trajectories.
  • trajectory-compiler : turn an agent audit log into long-context QA pairs.
  • loop-policy-update : outer-loop measurement for LLM-powered scoring policies.
  • ships : the public shipping log. The ritual, the rules, the examples.

More

workloft.ai · Labs · @alfred_workloft

Built in the open by Alfred Churchill.

Popular repositories Loading

  1. ships ships Public

    Workloft's public shipping log — the ritual, the rules, examples.

    Python

  2. trajectory-compiler trajectory-compiler Public

    Turn an agent audit log into long-context QA pairs. ACC applied to production trajectories. Open under MIT.

    Python

  3. loop-pilot loop-pilot Public

    A todo system the agent cannot cheat. Claude Code Stop hook + TTL escalator + hardened snooze + self-hosted dead-man. The Workloft Loop watertight pilot.

    Python

  4. civiclaw civiclaw Public

    Open-source, audit-native agent runtime for UK public sector (DSAR, FOI, EIR, EU AI Act Annex IV + FRIA). Mirror of gitlab.com/Alfpl/civiclaw.

    Python

  5. auto-rubrics auto-rubrics Public

    Auto-generate evaluation rubrics from agent audit-log trajectories (PhoneWorld pattern applied to action logs)

    Python

  6. loop-policy-update loop-policy-update Public

    Outer-loop measurement for LLM-powered scoring policies (two-level autoresearch)

    Python

Repositories

Showing 7 of 7 repositories

People

This organization has no public members. You must be a member to see who’s a part of this organization.

Top languages

Loading…

Most used topics

Loading…