Frontier Intelligence Workflows

Transform emerging-technology claims into evidence records, governance review, and bounded release decisions.

Frontier Intelligence Workflows is an MIT-licensed public repository template for teams working where claims move faster than validation.

It helps teams preserve sources, separate observation from inference, document uncertainty, record human review, and decide what may responsibly be released. Use FIW when a public-facing claim needs traceable evidence, explicit uncertainty, human review, and a bounded release decision.

Automation validates structure and integrity. It does not determine truth.

Start in five minutes

1. Read Start Here.
2. Review the completed synthetic example.
3. Copy the evidence card, AI governance gate, or Perception Integrity assessment.
4. Run the relevant evaluator and the repository checks below.
5. Record human review before making a release decision.

Run Perception Integrity on one assessment

mkdir -p profiles/perception-integrity/examples/FIW-SYN-ADOPTER ../fiw-validation
cp templates/perception-integrity-assessment.json \
  profiles/perception-integrity/examples/FIW-SYN-ADOPTER/PIA-FIW-SYN-ADOPTER.json
python scripts/run_perception_integrity.py \
  --root . \
  --assessment profiles/perception-integrity/examples/FIW-SYN-ADOPTER/PIA-FIW-SYN-ADOPTER.json \
  --evaluated-at 2026-08-08T12:00:00Z \
  --json-output ../fiw-validation/PIA-FIW-SYN-ADOPTER.validation.json

The output is advisory and always requires a human decision.

Adapting this template

Adding an approved file intentionally changes the committed manifests. The owner of a fork is its maintainer and must refresh metadata before full validation:

PYTHONDONTWRITEBYTECODE=1 python scripts/refresh_release_metadata.py --root .

Then run the complete read-only gate.

What this produces

• source-linked evidence records
• explicit observation, inference, confidence, and limitation fields
• governance and AI-assistance review
• decision-ready briefs
• human-approved release decisions
• deterministic manifests and release packages

What this does not do

This repository does not certify claims, prove technical performance, qualify suppliers, authorize deployment, replace expert review, or establish legal or regulatory compliance.

Core workflow

Claim
  ↓
Source and evidence record
  ↓
Evidence card
  ↓
Governance review
  ↓
Optional Perception Integrity assessment
  ↓
Human decision
  ↓
Bounded release decision

A claim may move forward only within the scope supported by its evidence and documented review.

Completed public-safe example

The fictional case FIW-SYN-001 follows an advanced-component readiness claim from source record through release decision.

Its result is intentionally bounded: the evidence supports a limited evaluation, not qualification, certification, production readiness, or public performance claims.

Repository map

.github/workflows/       Enforced validation and release-package checks
docs/                    Evidence, governance, assurance, and operations records
examples/                Completed public-safe examples
profiles/                Optional deterministic profiles, including Perception Integrity
scripts/                 Repository validation, evaluation, and deterministic packaging
templates/               Reusable evidence, governance, brief, and assessment templates
tests/                   Adversarial, release, and adopter UAT coverage

Validation

Prerequisites: Git, Python 3, a Bash-compatible shell, and sha256sum or an equivalent SHA-256 verification tool.

mkdir -p ../fiw-validation ../fiw-release
python scripts/run_tests.py --root . --json-output ../fiw-validation/source-test-summary.json
python scripts/compile_sources.py --root .
python scripts/validate_repo.py --root . --json-output ../fiw-validation/validation-report.json
python scripts/build_release.py --root . --output-dir ../fiw-release --check
sha256sum -c MANIFEST.sha256
git diff --check
test -z "$(git status --porcelain)"

Validation checks structure, active identity, version consistency, links, public/private boundaries, common secret patterns, intake consistency, example traceability, manifests, and workflow controls. A passing check is evidence only for the control it tests.

Public discipline boundary

This public repository may contain templates, workflows, synthetic examples, public-safe evidence records, and release evidence. It must not contain private client work, real supplier intelligence, proprietary scoring logic, patent strategy, controlled information, credentials, operational vulnerabilities, or unsupported claims presented as validated truth.

See IP Boundary, Security, and Limitations.

AI-assisted work

AI may assist with drafting, organization, summarization, and consistency checking. Meaningful AI assistance must be disclosed and reviewed under AI Provenance. Human judgment remains final.

License and brand boundary

Repository code, templates, documentation, and workflows are provided under the MIT License unless otherwise noted. Bridge Node 7 names, marks, website copy, visual identity, and protected website content are not granted under this license. Use of this repository does not imply Bridge Node 7 endorsement.

Published by Bridge Node 7.

Naming history

Historical releases through v0.1.3 used the name “Emerging Tech Proof OS.” The current public name is Frontier Intelligence Workflows.

Status

v0.3.0 - Adopter Enablement

This release adds an adopter-facing Perception Integrity runner, a schema-valid assessment template, a documented metadata-refresh loop, normalized semantic comparison, stronger baseline secret-pattern coverage, consistent restricted issue intake, and direct UAT coverage. The Perception Integrity contract remains deterministic, advisory, and human-review-required.