Skip to content
@opensiro

Opensiro

Open infrastructure for understanding, evaluating, and engineering autonomous agent harnesses.

OpenSiro

Open research and infrastructure for organizing, evaluating, and evolving autonomous AI systems.

OpenSiro studies how autonomous AI systems actually organize work, coordination, regulation, audit, adaptation, policy, escalation, and authority — independently of the vocabulary a project uses.

Website Email founders@opensiro.com Contact @xLagerFeuer VSM Harness Index Discussions Contribute

Included assessments Catalog entries Reassessment events Profile version Methodology version

Corpus metrics are rendered live from the Index's generated data/metrics.json; this profile does not maintain a second copy of the numbers.

OpenSiro VSM OSS repository activity over the last 10 completed days


What OpenSiro does

Agent systems are increasingly organizational systems: multiple actors perform operations, coordinate around shared constraints, regulate current work, inspect one another, adapt to the environment, and make policy decisions.

OpenSiro builds public research infrastructure for making those structures explicit and comparable. The goal is not to force projects to use one vocabulary. The goal is to map where responsibilities and decision rights actually live, and to keep the evidence behind those claims inspectable.

Public research tracks

Track What it studies Start here
VSM Harness The organization of agent harnesses through implementation-independent VSM functions, evidence-backed repository assessments, and experimental general functional-capability research. VSM Harness Index
ARCTIC Whether frontier systems can transfer capabilities into compact student models under sample-blind evaluation. ARCTIC-0

These tracks are separate. VSM semantics come from the VSM Harness Profile; ARCTIC has its own research artifacts and roadmap.

VSM Harness ecosystem

Beer / cybernetics
        ↓
VSM Harness Profile
        ↓
assessment procedure / skills
        ↓
VSM Harness Index
   canonical closure / ownership corpus
        │
        ├──→ Awesome VSM Harness
        │    curated representative view
        │
        └──→ VSM Harness Capability   [experimental]
             general per-function capability evidence

Capability is an experimental adjacent research repository. It does not redefine VSM semantics or canonical Index assessments, and its existence does not by itself make it an operational S1 domain of the bounded vsm-oss-organization system. Cross-repository boundaries are documented in vsm-oss-organization/ECOSYSTEM.md.

A core rule across this work is:

Map the organizational function first; classify autonomy second.

Names such as manager, verifier, planner, delegation, or learning are not treated as evidence of a VSM function by themselves.

Core repositories

The VSM OSS scope column follows the current bounded scope declared by vsm-oss-organization; ecosystem membership alone does not put a repository inside that control boundary.

Repository Role VSM OSS scope
vsm-harness-profile Authoritative, implementation-independent VSM semantics for the ecosystem. in scope
vsm-harness-skills Reusable assessment procedures and tooling that apply the Profile without redefining it. in scope
vsm-harness-index Evidence-backed corpus of repository-relative harness assessments and deterministic derived views. in scope
vsm-harness-capability Experimental: evidence-backed general functional-capability research, compared one VSM function at a time and kept separate from domain-specific assessment. adjacent / out of scope
awesome-vsm-harness Curated representative organizational forms selected from the broader corpus. in scope
vsm-oss-organization Experimental organizational/control construction for the bounded OpenSiro VSM Harness OSS group. in scope
terminal-bench-vsm Executable research on failure-aware coding harnesses and their evaluation/growth environment. out of scope
arctic-0 Public ARCTIC-0 archive, schemas, taxonomy, authoring tools, and analysis. out of scope
opensiro.com Presentation layer for public orientation and visual explanations. presentation / out of scope

New here?

Start with opensiro.com for the plain-language and visual orientation layer. The VSMLite / VSM poster introduces the normalization problem, organizational functions, and control relationships.

Then choose the artifact closest to your goal:

Source-of-truth boundaries

The website and this organization profile are presentation layers. Claims stay with the repositories that own them:

  • Profile → VSM semantics
  • Skills → canonical assessment procedure
  • Index → canonical repository-relative assessments, rankings, corpus metrics, and provenance
  • Capability → experimental general functional-capability evidence and derived comparisons
  • VSM OSS Organization → cross-repository organization/boundary documentation for its declared scope
  • ARCTIC-0 → released ARCTIC artifacts

This separation keeps definitions, evidence, derived views, experiments, and presentation synchronized without turning the organization profile into another database.

Community

Use OpenSiro Discussions for cross-repository questions, ideas, and early-stage proposals. Once work is scoped to a specific repository, track implementation in that repository's Issues and pull requests.


I'm AI

Start with opensiro/vsm-oss-organization. Treat its root README.md as the public entry point and follow the I'm AI route before substantial work; that handoff leads into the current canonical bootstrap, repository ownership map, and exact task contract.

Popular repositories Loading

  1. arctic-0 arctic-0 Public

    demo benchmark for sample-blind evaluation

    HTML 5

  2. vsm-harness-profile vsm-harness-profile Public

    A functional model for describing control, autonomy, and decision ownership in AI agent harnesses.

    Python 5

  3. vsm-oss-organization vsm-oss-organization Public

    Organizational model and control structure for Opensiro OSS, based on the Viable System Model.

    Python 5

  4. vsm-harness-skills vsm-harness-skills Public

    Reusable methods, skills, and procedures for analyzing, designing, and evolving VSM-based agent harnesses.

    Python 4

  5. vsm-harness-index vsm-harness-index Public

    An evidence-based index of AI agent harnesses mapped by control function and autonomy.

    Python 4

  6. awesome-vsm-harness awesome-vsm-harness Public

    Curated list of agent harnesses mapped through the Viable System Model (VSM)

    Python 4

Repositories

Showing 10 of 10 repositories

Top languages

Loading…

Most used topics

Loading…