From 57cbd68b47a54ac97d7259d4265b7f5a9049cd1e Mon Sep 17 00:00:00 2001 From: Perseus Computing <51974392+tcconnally@users.noreply.github.com> Date: Mon, 17 Aug 2026 23:19:38 +0000 Subject: [PATCH] fix(government): make capability route single-purpose --- government/index.html | 86 ++++++++++++++++++++----------- tests/test_capability_download.py | 17 ++++++ 2 files changed, 72 insertions(+), 31 deletions(-) diff --git a/government/index.html b/government/index.html index 54c80c8b..ebf9852b 100644 --- a/government/index.html +++ b/government/index.html @@ -3,8 +3,8 @@ -Perseus Computing | Context, memory, and evidence inside the boundary - +Capability statement | Perseus Computing + @@ -12,44 +12,68 @@ - -
-
Defense and regulated deployment

The system around the model.

Perseus Computing builds live context, governed memory, and evidence infrastructure for AI agents operating in controlled, disconnected, private-VPC, and regulated environments.

Perseus resolves current state before an agent acts. Vault retains governed memory. Ledger records what happened and why it can be reviewed.

-
What we build
01 / context

Live context before action

Perseus resolves repository, ticket, deployment, and decision state into bounded context before an agent acts.

02 / memory

Governed memory after action

Perseus Vault retains encrypted, bitemporal memory with durable journaling and explicit authority boundaries.

03 / evidence

Reviewable provenance

Ledger links actions and decisions to evidence, authority, and time instead of leaving the record in an opaque dashboard.

-
Why the boundary matters

AI can produce a plausible answer while losing the state that made it useful.

Perseus keeps the working context, memory, and evidence chain in the same boundary as the workflow. The deployment can be on-premises, air-gapped, in a private VPC, or on a controlled network with no required cloud service, API key, or vendor runtime.

-
Where it fits
Cyber & Networks

AI/ML, DevSecOps, cyber, and enterprise IT workflows where context and evidence must stay inside the controlled network.

Start here
C3BM

Software factories and distributed decision workflows; enabling infrastructure, not a claim to own the mission platform.

Enabling fit
Electronic Systems

Partner-led support where local, reviewable AI connects to sensing, EW, PNT, or mission-system workflows.

Partner-led
-
Continue the conversation

Bring the workflow. We will bring the smallest credible path.

A useful first discussion has a mission workflow, a data boundary, and an integration surface. The capability statement keeps the rest concise.

+ +
+
Government · primes · technical partners

Capability statement.

Infrastructure for AI agents that need their context, memory, and evidence to stay inside the mission boundary.

Perseus Computing builds local-first infrastructure for controlled, disconnected, private-VPC, and regulated environments. Start with the document; choose a tailored version only when the conversation calls for it.

+
Three send-ready versions

Choose the version for the conversation.

Each version keeps the same evidence and procurement facts while changing the front door and fit language.

Primary fit

Cyber & Networks

For AI/ML, DevSecOps, cyber, and enterprise IT workflows where context, memory, and evidence must stay inside the controlled network.Cyber / IT

Download PDF
Enabling fit

C3BM

For software factories and distributed decision workflows where Perseus can sit beside—not replace—the mission platform.Software factory / decisions

Download PDF
Partner-led

Electronic Systems

For a prime, integrator, or technical partner exploring local, reviewable AI around sensing, EW, PNT, or mission-system workflows.Partner / integration

Download PDF
+
What is inside

A reader can scan it in two minutes.

Page 01 / fit

What Perseus builds

The problem with lost state, the local context–memory–evidence system, the audience fit, and the smallest useful first conversation.

Page 02 / proof

Why the boundary matters

Measured evidence, procurement identifiers, licensing and SBOM posture, deployment options, and a direct contact path.

+
At a glance

The facts a buyer needs first.

Deployment
Local-first

Air-gapped, on-premises, private VPC, or controlled network.

Assessment
110 / 110

SPRS score; CMMC Level 2 self-assessment posture is scoped explicitly.

Identity
CAGE 22JC5

Perseus Computing LLC · UEI PJS2LW7HAK35.

License
MIT

SBOM materials published; no mandatory proprietary runtime.

+
Need the right front door?

Bring the workflow. We will bring the smallest credible path.

Send the mission workflow, its data boundary, and the integration surface. Thomas can point you to the right version or technical conversation.

- + diff --git a/tests/test_capability_download.py b/tests/test_capability_download.py index 17728cdc..492e6e8a 100644 --- a/tests/test_capability_download.py +++ b/tests/test_capability_download.py @@ -43,3 +43,20 @@ def test_tailored_capability_artifacts_exist_and_public_copy_is_not_internal_wor "intentionally short enough", ): assert phrase not in public_copy + + +def test_government_landing_is_single_purpose_and_exposes_all_three_custom_routes(): + landing = (ROOT / "government" / "index.html").read_text(encoding="utf-8") + assert "Capability" in landing and "statement." in landing + assert "Choose the version for the conversation." in landing + assert "Cyber-Networks.pdf" in landing + assert "C3BM.pdf" in landing + assert "Electronic-Systems.pdf" in landing + for retired_marker in ( + "Deploy where the cloud cannot reach.", + "Security posture", + "Deployment models", + "Continue the review", + "Sovereignty", + ): + assert retired_marker not in landing