METATROPOLOUS.COMResearch-led technology · Scoped commercial services
Meta GuideGitHub & methods

PUBLIC GITHUB & TECHNICAL CONTEXT

Show the sources.
Explain the evidence.

This page links to the publicly available upstream software used as technical context and explains how Meta Guide evaluations are reported. Private project CI artifacts are not exposed.

WHAT THESE LINKS ESTABLISH

Public source context
is not a product certificate.

The repositories below allow visitors to inspect the upstream open-source projects named in Meta Guide's genuine defect investigations. They do not contain Meta Guide's private trial receipts, nor do they imply endorsement by upstream maintainers.

Read evaluated Meta Guide outcomes ↗

SELECTED RESULTS FROM GITHUB-HOSTED EVALUATIONS

What our own testing
has actually established.

The following are manually curated, non-proprietary summaries of recorded development evaluations. They are not live repository widgets, public CI logs or independently published third-party certifications.

8/8Controlled causal-repair ladderAll eight cases passed through five levels, including four required corrections at Level 4.
4/4Nested controlled repair cohortFirst four of the same eight cases; passed at all five levels.
28/28Targeted historical test replaysThree source-pinned suites in recorded independent environments.
6/6Top-three localization matchesSix assessable authentic cases; five matched the maintainer-changed file at rank one.
8,415Historical CI run identities reconciledRun-level metadata, not 8,415 individual passing test assertions.
1Authentic upstream software repairQuart, with separate holdout, recovery and rollback evidence.

Private signed artifacts remain inaccessible to public visitors. Publication of independently reviewable redacted receipts would be a separate governed release, not an automatic feed from a private repository.

OPEN-SOURCE CONTEXT

Explore the projects
behind the cases.

The links go only to verified public upstream repositories. No private repository or GitHub Actions archive is linked.

PUBLIC UPSTREAM REPOSITORY

Quart

An asynchronous Python web framework. A single distinct Quart defect was successfully repaired in Meta Guide's private recorded evaluation, including a sealed holdout and rollback.

View upstream Quart on GitHub ↗
PUBLIC UPSTREAM REPOSITORY

FastAPI

An open-source framework represented in frozen localization and repair investigations. Being part of a test cohort does not mean its historical defect was solved by Meta Guide.

View upstream FastAPI on GitHub ↗
PUBLIC UPSTREAM REPOSITORY

Sanic

A Python asynchronous web server/framework included in the recorded authentic upstream defect cohort. The revised localization system ranked the changed file within the top three in the controlled study.

View upstream Sanic on GitHub ↗
PUBLIC UPSTREAM REPOSITORY

Luigi

A workflow-orchestration project represented in genuine code-localization and repair experiments. Source-relevance investigation results were compared against independent maintainer changes; this is a localization reference, not a repair claim.

View upstream Luigi on GitHub ↗
PUBLIC UPSTREAM REPOSITORY

Flask

An open-source Python web framework evaluated in the larger defect cohort. Its historical upstream case provides independent context for source localization and observed test behavior.

View upstream Flask on GitHub ↗
PUBLIC METHODS REFERENCE

GitHub Actions artifacts

Learn how CI artifacts and access permissions work. Private GitHub Actions artifacts are not automatically available to public visitors, so they cannot substitute for an approved public evidence brief.

Read official GitHub documentation ↗

HOW TO READ THE NUMBERS

Six distinct
evidence categories.

A result is meaningful only if its denominator, mechanism, evaluation environment and scope remain attached.

01

Source ancestry

Records commits, blobs and source identities. This establishes traceability, not software correctness.

02

Test archive reconstruction

Counts CI run identities and historical metadata without presenting run totals as independent passing tests.

03

Localization

Measures whether proposed source files overlap with later independently observed maintainer-edited files.

04

Repair verification

Requires a reproduced failing baseline, candidate, separate test/holdout results and preserved originals.

05

Governance checks

Measures whether specified protected behavior and authority are actually enforced under bounded conditions.

06

Generalization

Needs newly unfamiliar cases with independent scoring. Broader reliability is still being evaluated separately from the controlled repair ladder.

Publication boundary: Meta Guide's operational repository and audit receipts are private. Public links on this page refer only to upstream open-source projects or public methodology documentation. A sanitized technical evidence pack may be prepared for prospective customers after owner approval; none is represented here as a published third-party certification.

Independent verification: Public project listings verify the origin and nature of the upstream code, not Meta Guide's performance. The Evidence page presents recorded outcomes with limitations.

EVIDENCE-LED PILOTS

Want to examine a result
relevant to your system?

Request a scoped technical briefing. We can discuss which non-proprietary findings support your case.

Request a technical briefing