Skip to main content

pgs_runtime

Deterministic execution engine for Protocol-Governed Systems.

The runtime traverses a precompiled execution graph and produces traceable, governed outcomes.
It does not discover behavior. It does not interpret intent. It does not contain business logic.

Behavior is declared in protocol, executed by runtime, implemented in capabilities, and observed via traces and state.

New to PGS? This is one of eight repositories in the Protocol-Governed Systems ecosystem. For orientation, architecture overview, and end-to-end execution, start at pgs_workspace.


What this component is (and is not)

This is:

  • A generic execution substrate
  • A graph traversal engine
  • A trace generator
  • A host for capability implementations (CT/CS bindings)

This is not:

  • A workflow authoring system
  • A rules engine
  • A business logic container
  • A framework with pluggable behavior

All behavior is declared and compiled before execution. The runtime only consumes a sealed snapshot.


Inputs → Outputs

Inputs:

protocol_snapshot/   ← compiled artifacts (from pgs_compiler)
payload              ← external input (JSON)
data_root            ← state storage boundary
workspace            ← execution context

Outputs:

traces/<TRACE_ID>/
├── <TRACE_ID>.jsonl   ← append-only execution log
├── <TRACE_ID>.md      ← human-readable summary
└── <TRACE_ID>.png     ← visual execution path

data/
├── registry/          ← idempotent state
├── events/            ← append-only history
└── ...                ← additional declared side-effects

CLI surface

omnibachi run \
  --wf blockchain::WF_REGISTER_ACTOR_UNVERIFIED_V0 \
  --payload payload.json \
  --data-root ./data \
  --workspace .

omnibachi examine ./traces/<TRACE_ID>/<TRACE_ID>.jsonl

That is the entire public interface. Everything else is governed by the protocol snapshot.


How execution works

You run a workflow. The runtime:

  1. Loads the compiled DAG from protocol_snapshot/
  2. Resolves inputs via JSONPath expressions
  3. Executes CT/CS nodes (transforms and side-effects)
  4. Routes based on declared outcomes
  5. Emits trace + state

No logic exists in the runtime for any specific workflow.

Example: omnibachi run --wf blockchain::WF_REGISTER_ACTOR_UNVERIFIED_V0 ...

The runtime has no knowledge of blockchain. It only traverses the precomputed graph compiled from that workflow's protocol declaration.


What makes this runtime different

1. Compile-time closure

All routing, bindings, and side-effect semantics are resolved before execution begins:

  • no reflection
  • no dynamic imports
  • no fallback logic

Execution is a pure traversal of a sealed graph.
No runtime decision can alter the graph.

2. Outcome-driven control flow

Every node produces an explicit outcome:

SUCCESS
ALREADY_EXISTS
VIOLATION
...

These outcomes map to named edges in the graph. There is no implicit branching, no hidden control flow, no exception-driven logic.

3. Side-effects are governed, not coded

Side-effects (CS) declare their own semantics:

Behavior Example
Idempotent registry insert → ALREADY_EXISTS on repeat
Append-only event stream → always records
Mutable state overwrite (explicitly declared)

The runtime executes these semantics as declared. It does not interpret them.

4. Execution ≠ effect

A workflow can succeed even when a side-effect returns ALREADY_EXISTS. Execution is structural. Effects are governed independently. This separation is visible in traces and state.

5. Trace is first-class output

Every run produces a complete causal execution record keyed by trace_id. This enables:

  • deterministic replay
  • auditability
  • debugging without code instrumentation

Where this fits in the system

PGS is intentionally split into layers:

Layer Repo Responsibility
Governance pgs_governance Invariants, constitutional rules, structural definitions
Compilation pgs_compiler Produce protocol_snapshot/
Transport pgs_transport Ingress/egress adapters for HTTP and CLI
Execution pgs_runtime ← here Traverse compiled graph deterministically
Capabilities pgs_capabilities Provide CT/CS implementations
Domains pgs_blockchain, pgs_ai_governance Real-world workflows
Entry point pgs_workspace Run and observe

What you should explore next

Go here To...
pgs_workspace Run end-to-end and observe traces
protocol_snapshot/ Inspect what the runtime actually reads
pgs_governance + pgs_compiler Author and compile new workflows
pgs_capabilities Add new capability implementations

Research context

This implementation demonstrates:

"Extensibility by declaration, not refactor."

And more broadly: governance-driven execution, compile-time closure of behavior, separation of execution from semantics.


Final note

The runtime does not decide what to do.
It only executes what has already been decided.

If you find yourself wanting to add logic to this runtime — stop.
That logic belongs in the protocol (declaration) or in capability implementations (CT/CS).
The runtime should remain boringly deterministic.


License

Apache-2.0. See LICENSE and NOTICE for details.

Metadata

Release files for omnibachi 0.2.1

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for omnibachi 0.2.1
File Size Uploaded Provenance
omnibachi-0.2.1.tar.gz 63.0 kB Without provenance Details

Built distribution (wheel)

Table of built distributions (wheels) for omnibachi 0.2.1
File Interpreter ABI Platform
omnibachi-0.2.1-py3-none-any.whl Python 3 none any Details

Total release size: 144.8 kB

Release files / omnibachi-0.2.1.tar.gz

Download URL omnibachi-0.2.1.tar.gz
Size 63.0 kB
Tags Source
SHA-256 checksum
How to use checksums
26adf17470f10974e64374fb785f439ad83a88ab9f9f158388f5fd8de85d687b
BLAKE2b-256 checksum
How to use checksums
68c4a2c3a3bcaf3586202d029b378f7ba13d6fa6829a02de23f64a510055b7e6
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.10.11

Release files / omnibachi-0.2.1-py3-none-any.whl

Download URL omnibachi-0.2.1-py3-none-any.whl
Size 81.9 kB
Tags Python 3
SHA-256 checksum
How to use checksums
393cbc20ba65bf2306281782ae276088fe9ceaf80b49ca075f6d2309c04376ac
BLAKE2b-256 checksum
How to use checksums
2330214ff3a530bfbfb303a7f32153d917a1d52e1e7a1d300c8dc30b9d799eac
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.10.11

Security for omnibachi 0.2.1

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page