Signal Room

Inspectable decision support for choosing AI models, configurations, agent harnesses, and self-hosting options.

---

Note

Applied research, actively maintained — and intentionally non-exhaustive. Pricing changes, vendor-reported claims and heterogeneous benchmarks are not universally comparable, and its recommendations may not fit every workload. Evidence classes, source links, and measurement caveats stay visible so you can make your own tradeoffs.

Full detail: Project Status.

---

Site: https://projectious-work.github.io/ai-market-research/ Signal Room report: https://projectious-work.github.io/ai-market-research/report/ Latest release:

What this is

Choosing a model, a reasoning configuration, an agent harness, or a self-hosting setup means comparing vendor claims that are rarely apples to apples. Signal Room is a non-exhaustive, inspectable report: every number is either measured, vendor-reported, or explicitly marked unknown, and every claim keeps a canonical source link.

It tracks the current model roster (Anthropic, OpenAI, Google, xAI, Meta, Mistral, Cursor, and major Chinese labs) with provider-native reasoning configurations, separate TTFT/throughput/task-latency speed evidence, a quota-burn cross-matrix, agent harnesses (Claude Code, Codex CLI, OpenCode, and others), and dated self-hosting economics (GPU pricing, local Mac fit, quantization). See Data Methodology for the full breakdown, formulas, and evidence classes.

How it's built

• data/*.json — the roster, model configurations, and normalized report metrics that feed the report.
• src/dashboard.template.html + src/scripts/build.py — validate and embed the JSON inputs into a single self-contained dist/dashboard.html. No JavaScript framework, build chain, or runtime CDN dependency.
• content/ assets/ layouts/ static/ — the Hugo + Docsy documentation site, styled from the projectious.work brand system. The report is embedded full-bleed at /report/.

Research-data rules (source rights, attribution, retention, verification, privacy review) are documented in Research Data Policy.

Quick start

Requires python3, bash, git, and (for deploys) the gh CLI. Building the documentation site additionally requires hugo (extended) and node/npm.

bash src/scripts/build.sh           # build the report: dist/dashboard.html
bash src/scripts/release-check.sh   # validate + rebuild + sanity-check
bash docs/scripts/serve-docs.sh     # serve the Hugo/Docsy site, report at /report/

Full walkthrough: Quick Start.

Deploy

GitHub Pages is fed from the root of the gh-pages branch by a local script. No GitHub Actions workflow is used or permitted. Architecture: DEC-20260727_0742-ClearSpring (supersedes DEC-20260517_1455-DeftLynx).

bash src/scripts/release-check.sh
bash docs/scripts/deploy-docs.sh --message "deploy: refresh signal room"

The deploy script rebuilds the report and the Hugo site by default, stages the payload in a temporary Git worktree, pushes gh-pages without force, and verifies that Pages uses the legacy branch source with HTTPS.

Release

Releases follow the repository's phased release process, defined in AGENTS.md:

bash src/scripts/release.sh --list
bash src/scripts/release.sh --phase 0

Do not cut a version directly before the release gates have been evaluated.

Repository layout

data/                          Source JSON, normalized metrics, archives
src/                           Report template, build script, briefing prompt
dist/                          Built dashboard.html (gitignored)
content/ assets/ layouts/      Hugo + Docsy documentation site
docs/static/logo/              Signal Room mark (light/dark/mono variants)
scripts/{build,serve,deploy}-docs.sh   Hugo site build / serve / publish
docs/themes/docsy/             Docsy theme (git submodule)
context/                       processkit project context (decisions, logs, …)
AGENTS.md                      Provider-neutral agent instructions

Contributing

This is a public, executive-oriented market report and a reusable static dashboard implementation. The code is MIT-licensed; fork it freely if the structure is useful for your own market-watching workflow.

See CONTRIBUTING.md for validation, canonical Git identity, and deployment requirements. Security reports follow SECURITY.md.

Project status

Lifecycle: active. The market roster and published report are actively maintained. Historical snapshots remain available for provenance, but only the current report and latest tagged release are supported.

License

Unless otherwise noted, the copyright holder grants the MIT License for all versions of this repository, including historical commits and tags. The full license text is in LICENSE. © 2026 projectious.

Brand and design system © projectious.work. The Signal Room mark is derived from that system.