From 3bc7b6536fe468c873d7fd36fef8fc7406cc5baa Mon Sep 17 00:00:00 2001
From: Claude
Date: Thu, 30 Jul 2026 02:37:06 +0000
Subject: [PATCH 1/3] docs: ecosystem slide deck + profile, vertical, audience,
and property analyses
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Adds docs/deck/: a self-contained 22-slide HTML deck (site dark canon,
two-register narrative) presenting every engine alone and as the system,
plus four analysis documents — the 159-repo profile census, value maps for
the sales/marketing and provenance/privacy verticals, the target-audience
matrix, and the two-property (site + Flywheel surface) assessment with
recommendations.
Docs-only change: no Dart sources touched, so the flutter analyze/test gates
do not apply (toolchain also unavailable in this environment). The deck HTML
is a single self-contained artifact; the sub-300-line gate is held for the
markdown files.
Co-Authored-By: Claude Fable 5
Claude-Session: https://claude.ai/code/session_01PbFKzRRinxkZ34EwoA8Jwf
---
docs/deck/01-github-profile-analysis.md | 119 +++++
docs/deck/02-vertical-value.md | 82 ++++
docs/deck/03-target-audiences.md | 60 +++
docs/deck/04-property-assessment.md | 120 +++++
docs/deck/flywheel-telos-deck.html | 599 ++++++++++++++++++++++++
5 files changed, 980 insertions(+)
create mode 100644 docs/deck/01-github-profile-analysis.md
create mode 100644 docs/deck/02-vertical-value.md
create mode 100644 docs/deck/03-target-audiences.md
create mode 100644 docs/deck/04-property-assessment.md
create mode 100644 docs/deck/flywheel-telos-deck.html
diff --git a/docs/deck/01-github-profile-analysis.md b/docs/deck/01-github-profile-analysis.md
new file mode 100644
index 0000000..cd8cbb5
--- /dev/null
+++ b/docs/deck/01-github-profile-analysis.md
@@ -0,0 +1,119 @@
+# GitHub profile analysis — HarperZ9 / ZentropyLabs
+
+> Census taken 2026-07-30 from the live account listing. Counts are exact at
+> that timestamp. Companion files: 02-vertical-value.md, 03-target-audiences.md,
+> 04-property-assessment.md. Slide deck: flywheel-telos-deck.html.
+
+## Headline numbers
+
+| Measure | Count | Note |
+|---|---|---|
+| Total repositories | 159 | 158 under HarperZ9 + ZentropyLabs-ai/.github |
+| Original (non-fork) | 104 | the portfolio proper |
+| Forks | 55 | upstream OSS + curated awesome-lists + eval/agent tooling |
+| Private | 37 | security line, dev monorepos, site experiments |
+| Public | 122 | |
+
+## The five-tier taxonomy (per the site's own catalog)
+
+### Tier 1 — the fourteen public engines
+flywheel (route + verify; desktop client v0.2.2), telos (the shared workbench,
+live v0.2.0), index (2.9.0 beta, PyPI `index-graph`), gather (1.6.1 shipped,
+PyPI `gather-engine`), forum (1.13.0 shipped, PyPI `forum-engine`), crucible
+(1.2.0, PyPI `crucible-bench`), emet (1.1.0, PyPI `emet`, four independent
+language implementations), buildlang (v1.2.0, 1,521 passing tests), learn
+(v1.6.0), relay, plexus, mneme, studio-engine, build-color (1.0.2 beta,
+PyPI `build-color`).
+
+Six confirmed PyPI packages — the site publicly corrected an earlier "19"
+over-count to this honest figure.
+
+### Tier 2 — the private line ("public where safe")
+- **orca** (Runtime) — security-engagement workbench, v1.0.0, 361 tests;
+ metadata-only, cannot reach the target it assesses. The most revenue-shaped
+ private asset.
+- **aleph** (Checkpoint) — release/supply-chain coordination across docs,
+ contracts, MCP declarations, CI, receipts.
+- **kun** (Vault) — path-only receipts, rotation discipline, value-free.
+- **behavior-transform.io** (Boundary) — public repo; mode-aware
+ read/write/exec/fetch/input receipts; raw content stays in local adapters.
+- **seed** (Lab) and **sofer** (Ledger) — private/proprietary; honestly labeled
+ as needing a public-safe split (credential-pattern corpora inside).
+- Adjacent private security cluster: offensive-platform, red-team-platform,
+ adversarial, protected-corridors, protected-research, legal-intl,
+ warden-reporting / warden-clp / warden-prefire-primitive (warden as a public
+ brand is retired into proof-surface), emet-internal, met-monorepo,
+ ai-safety-prefire, ai-safety-guardrail-manager, unified-engine, aurora,
+ apps, flywheel-host.
+
+### Tier 3 — substrates
+proof-surface (652 tests, zero deps — the shared proof-packet foundation),
+coherence-membrane, accountable-surface, accountable-engine (Forming; "the
+inward half is owed"), provenance-sensorium, reconcile, raw (D3D11 mid-frame
+graphics proxy), witnessing-spine (five regulated-sector adversarial
+steelmans, DOI 10.5281/zenodo.20778927), senses-and-sensibility,
+faithful-transpile, coherence and kernels lines (signal-kernels,
+anomaly-kernels — C++23 header-only).
+
+### Tier 4 — bricks (release + agent toolkit)
+secret-redact-io, release-surface-scanner, public-surface-sweeper,
+repo-proof-index, proof-surface-report, model-provenance-validator,
+gpu-trace-validator, agent-audit, agent-routing-kit, agent-hook-pack,
+context-curator-lite, workflow-harness-lite, consulting-template-kit,
+calibrate-pro, wol-pi.
+
+### Tier 5 — research + receipts-in-the-wild
+Receipt repos aimed at other ecosystems: crewai-kickoff-replay-receipt,
+vllm-ci-forum-receipt. Research surface lives on the site: 3 theses,
+10 experiment packets, 6 DOI'd preprints (arXiv endorsement-pending),
+with a formal claim-status ladder (SOURCE_LEAD → … → LAW_CANDIDATE, plus
+UNVERIFIABLE).
+
+## The build/creative lines
+
+- **Build product line** (former "Quanta", renamed): build-engine, build-color,
+ build-oracle, build-ui, build-finance, build-universe, build-ecosystem,
+ buildlang + buildlang-vscode + buildlang-tmLanguage.
+- **Graphics/ENB line** (the consumer-traction record): elder-enb (900k+
+ downloads, 150k+ unique users, ~280 releases over ~2 years), truth-enb,
+ elder-weathers, enb-runtime-core, enb-validation-lab, enb-bridge-f4 (private),
+ brender-archival, skyrimbridge, skse/skse-playground (private), gaussian
+ splats (site page), engine-revival, studio-engine.
+- **Site/property repos**: HarperZ9.github.io (deployed output; canonical
+ source telos-v2, private), harpercompliance-site (private),
+ sendmyletter-site (private), freelance (private).
+
+## The fork belt (what it signals)
+
+55 forks cluster into: agent/eval infrastructure (pydantic-ai, deepeval,
+langfuse, inspector, agent-runtime-observatory, SmallHarness, runtrail,
+scorekeeper), MCP ecosystem curation (a dozen awesome-mcp-* lists, registry,
+python-sdk), core Python tooling (numpy, poetry, setuptools, isort, tomlkit,
+typeguard, grimp, datasette, llm), and agent-security reading
+(system_prompts_leaks, awesome-ai-agents-security). Read as a map: the
+account tracks exactly the frontier its products verify — agents, evals,
+MCP, and the supply chain under them.
+
+## Engine-state cross-check (dev vs public)
+
+| Surface | Routes / lanes | Version |
+|---|---|---|
+| `flywheel` public repo gateway | 11 API routes | 0.1.0 |
+| `local-model` dev engine gateway | 96 API routes | dev (CC-1 era) |
+| Lane registry (identical in both) | 7 lanes | gather 1.6.1 · crucible 1.2.0 · index 2.9.0 · forum 1.13.0 · learn 1.6.0 · telos 0.2.0 · local-model bundled |
+| Marketplace | 17 built-in MCP connectors + unlimited user-registered plugins | |
+| SUPERPROJECT roster | 11 mission-tier flagships (5 live MCP spine + 6 declared) | curated 2026-07-06 |
+
+The freshest engineering proof point is CC-1 "Certified Commons"
+(STATE.md 2026-07-25): `flywheel gate` reaches rewitness=MATCH from a fresh
+clone with byte-identical digests; shown able to FAIL first (verdict tamper →
+DRIFT); 85 + 243 passing tests on that slice; CI across three OSes.
+
+## Reading the whole profile in one line
+
+One person, one discipline — perceive, transform under a named criterion,
+carry a re-checkable receipt — applied across model infrastructure, compilers,
+color science, graphics, security practice, and research, with the honesty
+mechanics (published nulls, self-caught over-counts, maturity tiers) intact
+everywhere. The portfolio's weakness is not depth or range; it is that no
+single surface assembles this into an offer a buyer can say yes to.
diff --git a/docs/deck/02-vertical-value.md b/docs/deck/02-vertical-value.md
new file mode 100644
index 0000000..0de40a2
--- /dev/null
+++ b/docs/deck/02-vertical-value.md
@@ -0,0 +1,82 @@
+# Vertical value assessment — two industries, each tool alone and as the system
+
+> Companion to 01-github-profile-analysis.md. The claim discipline of the
+> portfolio applies to this document: where a value statement is a projection
+> rather than an observed result, it says so.
+
+## The shared premise (why either vertical pays)
+
+Machines made claiming nearly free; checking stayed expensive. Both verticals
+below are drowning in AI-produced output whose acceptance currently rests on
+confidence. The system's one operation — propose, verify against an external
+criterion, keep a re-checkable receipt — is the missing accept path. The
+pitch deck already names the wedge categories (originality-verification,
+accountable-agent-action, provenance) and is honest that they have no analyst
+TAM yet: that is the whitespace, and also the risk.
+
+## Vertical A — sales & marketing (greenfield)
+
+Grep-verified: the current site has zero sales/marketing surface area. No
+positioning debt, no incumbent framing to unwind — and no existing proof of
+demand either. Everything below is a mapping of shipped capabilities onto
+known agency/martech pains, not observed traction.
+
+| Asset | Alone, it is | In this vertical it becomes |
+|---|---|---|
+| **gather** + snapshot (citation freezing) | research intake with provenance receipts | Claims substantiation: every public sentence traceable to a frozen, hashed source that outlives the live web. Ad-claim defense files that assemble themselves. |
+| **forum** | agent orchestration on a replayable causal hash-chain ledger with human gates | Agency accountability: campaign automation the client can re-check instead of trusting a report. A deliverable format no incumbent offers. |
+| **flywheel** routing + receipts | one request shape over every provider, observed per-provider scoreboard | AI-spend governance: cost-honest routing, escalation receipts, the first "what did the model spend buy us" ledger. |
+| **studio-engine** + telos creative | seeded, replayable generative worlds/brand kits | Reproducible brand systems: same seed, same artifact — campaign creative that is re-derivable, auditable, and provably original-to-seed. |
+| **learn** | witnessed courses; every graded step writes a receipt | Enablement proof: onboarding and certification that produce receipts, not completion checkboxes. |
+| **mneme** | agent memory where every memory carries provenance | Account knowledge that survives turnover: "who told us this, when, from which document" as a queryable property. |
+| **crucible** | falsifiable claim verification | Marketing-claims QA and competitor-claim testing: register the claim, measure it, publish the verdict packet. |
+
+**As the system**: an agency or in-house team runs intake (gather) → brief
+verification (crucible) → campaign orchestration (forum) → creative (studio)
+→ delivery with receipts (flywheel/desktop), and hands the client one chained
+record. The sale is not "AI tools"; it is *defensible delivery*.
+
+**Honest boundaries**: no marketing-specific integrations exist today (no CRM,
+no ad-platform connectors — though 17 built-in MCP connectors plus the plugin
+registry are the extension path). First proof would need one design-partner
+agency.
+
+## Vertical B — provenance-critical, privacy-bound industries (assembled, not built)
+
+The material for this vertical already exists across seven site pages and
+several repos; it has never been assembled into one buyer-facing argument.
+
+| Sector | Live requirement | Shipped answer |
+|---|---|---|
+| **Compliance / model risk (finance)** | Regulators require model origin and lineage to be auditable; incumbent platforms record lineage as editable database rows (site: writing.html; witnessing-spine, DOI'd, steelmans five financial-sector cases). | Merkle-logged receipts with offline inclusion proofs; hash-chained verifiable store with audit tail; attestation binding sign-off to exactly what was reviewed, overclaims tracked. |
+| **Legal / e-discovery** | "The summary is not the record." AI summaries entering matters without their sources. | Citation freezing (page bytes fetched, hashed, stored), content-addressed corpora, comprehension gating (a teach-back that pasting the diff cannot pass). |
+| **Healthcare administration** | Summaries must keep uncertainty attached (field-guide clinical lane). | Honest nulls as first-class UI; UNVERIFIABLE rendered as loudly as a win; escalation with the failed local attempt on record. |
+| **Newsrooms / media** | Every public sentence must find its source (field-guide media lane). | gather → forum → crucible chain; per-block source hashes; tamper caught on re-read (demonstrated in the recorded demos). |
+| **Security / red teams (gov, law enforcement, AI labs)** | Authorized adversarial work needs records that hold up afterward, and a lawful-basis gate. | The existing security.html practice + ORCA (v1.0.0, 361 tests, metadata-only engagement workbench). Already revenue-shaped. |
+| **Regulated AI operations** | Agent boundaries must be provable, credentials must not travel. | Boundary receipts (raw content never leaves local adapters), capability grants per run, presence-only credentials in the OS keychain, zero telemetry, fully local loop. |
+
+**The privacy story is architectural, not contractual** — this is the
+differentiator dense-workflow industries actually buy:
+- The desktop client's only base URL is 127.0.0.1:8799; no outbound endpoint
+ exists in the codebase.
+- Credential *values* never enter the app, the gateway refuses to accept them
+ through the marketplace, and the UI shows presence + source only.
+- Proofs verify offline: a third party recomputes the Merkle root from the
+ leaf with no network, no account, no vendor.
+- Write/exec are grants per run; a verify stage without an exec grant reports
+ UNVERIFIABLE rather than pretending.
+
+**As the system**: the same loop, pointed at regulated work — intake with
+receipts, judgment that fails closed, orchestration with human gates, sign-off
+bound to coverage, and an audit trail that a regulator, opposing counsel, or
+incident reviewer can re-derive without trusting the operator.
+
+## Value of each tool alone vs. the system
+
+Each lane is deliberately "a full product that also runs alone" — gather,
+crucible, index, forum, learn are individually pip-installable with their own
+receipts. Alone, each competes in a crowded category on speed/cost and honesty
+(the battle map publishes THEY_LEAD rows). Together they occupy compositions
+with no incumbent: benchmark-claim escrow, accountable-agent-action,
+receipt-carrying creative, the concurrent-desktop agent. The system premium is
+the moat; the standalone tools are the doors in.
diff --git a/docs/deck/03-target-audiences.md b/docs/deck/03-target-audiences.md
new file mode 100644
index 0000000..d800a2f
--- /dev/null
+++ b/docs/deck/03-target-audiences.md
@@ -0,0 +1,60 @@
+# Target audiences — where, how, and why these tools matter on today's frontier
+
+> The frontier condition in one line: autonomous systems now produce work
+> faster than any review lane absorbs it, provenance duties are hardening into
+> regulation, and public AI claims go unchecked by default. Every audience
+> below is someone for whom one of those three pressures is already a budget
+> line or a liability.
+
+## The three frontier pressures → the system's three answers
+
+| Pressure (where the frontier bites) | The system's answer |
+|---|---|
+| **Agents outran review** — autonomous work lands faster than humans check it | The verified accept path: an external oracle decides; models propose, oracles dispose; UNVERIFIABLE is a legal, visible outcome |
+| **Provenance became law** — lineage/origin duties moving into regulation across finance, health, media | Receipts as infrastructure: content-addressed, Merkle-proven, offline-verifiable, vendor-independent |
+| **Claiming became free** — capabilities, benchmarks, and originality asserted daily, checked never | Claim escrow: sealed verdict packets a stranger, competitor, or court can recompute |
+
+## Audience matrix
+
+| Who | Where they live | Why now | How they enter | Value to them | Value to us |
+|---|---|---|---|---|---|
+| **Agency ops lead** | Campaign delivery, client reporting | Clients audit AI-made work; agencies have no proof layer | forum demo: the replayable run ledger | Defensible delivery — the client re-checks instead of trusts | First design partner in the greenfield vertical |
+| **CMO / martech owner** | Brand, content, AI-spend budgets | AI spend unmeasured; claims risk rising | flywheel routing scoreboard + gather claim receipts | Cost + substantiation in one surface | The paying wedge account for vertical A |
+| **GRC / model-risk officer** | Banks, insurers, model inventories | Lineage regulation arrived; current audit trails are editable rows | Witnessing Spine findings + a receipts inclusion-proof demo | Tamper-evident lineage a regulator can re-derive | The regulated flagship reference |
+| **Legal ops / e-discovery lead** | Firms, corporate legal departments | AI summaries entering the record without their sources | snapshot (citation freezing) + attestation | Summaries bound to frozen records; sign-off bound to coverage | The dense-workflow proving ground |
+| **Newsroom standards editor** | Media, fact-check desks | Generated content floods intake; provenance IS the product | field-guide media lane: gather → forum → crucible | Source-checked publishing with per-block hashes | A public credibility case study |
+| **Red team / security lead** | Government, law enforcement, AI labs | AI attack-surface work needs lawful, replayable records | ORCA + the security.html engagement gate | Engagement records that hold up afterward | The already-revenue-shaped practice |
+| **Clinical admin / health-ops** | Hospital administration, payer ops | Summaries must keep uncertainty attached | field-guide clinical lane (gather + forum) | Honest-null summaries; UNVERIFIABLE stays visible | Entry into the highest-stakes provenance market |
+| **Builders & researchers** | OSS, eval/agent infrastructure | Already served — the audience the site speaks to today | PyPI packages, recorded demos, DOI'd papers | Tools + receipts + a falsifiable research program | Contributors, field testers, standing — not revenue |
+
+## Where each audience is reached
+
+- **Agencies / CMOs**: martech and agency-ops communities; the deck's Act II-A
+ is the first artifact that speaks to them — nothing on the site does yet.
+- **GRC / model risk**: the Witnessing Spine paper is citable; model-risk and
+ AI-governance forums; the "editable rows vs. Merkle proof" demo is the hook.
+- **Legal**: e-discovery and legal-ops conferences; the citation-freezing demo
+ is self-explanatory to anyone who has fought over a dead link in a record.
+- **Newsrooms**: the field guide's media lane is already written as a script
+ with a break test — publishable as-is.
+- **Security**: the practice exists with a lawful-basis gate; the audience is
+ reached by referral and credential, not marketing.
+- **Builders**: already arriving via PyPI, the demos, and the papers.
+
+## The honest split
+
+The audience that already responds (builders, researchers) is not the audience
+that pays. The audiences that would pay (compliance, legal, agencies, regulated
+ops) are named nowhere on the current public surface except behind the
+security page's authorization gate. Closing that gap is a positioning task,
+not an engineering one — the evidence is already written, scattered across
+seven pages.
+
+## Why this matters to each side (the deck's dual promise)
+
+For the owner: each tool is independently sellable (pip-installable, own
+receipts, own demo), so every audience above is reachable without waiting for
+platform adoption. For the audience: the system never asks to be trusted —
+every claim it makes about their work arrives with the means to re-check it,
+which is precisely the property their regulators, clients, editors, and
+courts are beginning to demand.
diff --git a/docs/deck/04-property-assessment.md b/docs/deck/04-property-assessment.md
new file mode 100644
index 0000000..c6ab8aa
--- /dev/null
+++ b/docs/deck/04-property-assessment.md
@@ -0,0 +1,120 @@
+# Web-property assessment — the two domains
+
+> "Both domains" per the owner's direction: (1) the harperz9.github.io
+> portfolio/Project Telos site, and (2) the Flywheel product surface
+> (public engine repo + flywheel-desktop + its distribution page).
+> Assessed 2026-07-30 from full clones. No registered custom domain exists
+> today; behavior-transform.io appears only as a GitHub repository name.
+
+## Property 1 — harperz9.github.io (Project Telos site)
+
+### What it is
+72 sitemap pages (87 HTML files), actively maintained (PR #89 merged
+2026-07-29), hand-authored on a shared design cascade, deployed from the
+private telos-v2 source with a verification gate.
+
+### Strengths (VERIFIED)
+- Near-100% metadata hygiene: title/description/OG/Twitter on 71 of 76 root
+ pages (the five exceptions are intentional redirect stubs); reproducible
+ 1200×630 social cards generated from a committed template.
+- Tested like a product: ~30 JS unit tests, 8 Python site-contract tests, a
+ link-check crawler, a visual contract test, CI.
+- Accessibility by rule: skip links, reduced-motion mandates, ARIA labels.
+- An honesty discipline with no known peer: a published self-audit
+ (SITE-FIX-LIST.md) that caught its own stale versions and over-counts, a
+ public correction note ("the honest figure was six"), and a published null
+ against its own product's uplift claim. All previously-listed defects were
+ re-verified fixed at assessment time.
+- Deep evidence library: 4 recorded receipt-backed demonstrations, 14 research
+ pages with named falsifiers, 6 DOI'd preprints, a 15-page investor deck.
+
+### Gaps (DRIFT)
+- **No custom domain.** Free github.io hosting; zero owned domain authority;
+ brand identity split three ways (Zain Dana Harper / Project Telos /
+ ZentropyLabs, with a ZentropyLabs-ai GitHub org besides).
+- **Zero analytics** (grep-verified: no GA/GTM/Plausible/Umami/Fathom/Clarity).
+ The site is unmeasurable — by its own thesis, it cannot make a measured
+ claim about itself.
+- **No conversion instrument.** No form, calendar, newsletter, or pricing;
+ every CTA terminates at a gmail address. The consulting CTA has no landing
+ page.
+- Homepage is a 4.6 KB JS-dependent shell that leads with a compiler test
+ count rather than a value proposition.
+- Residue: 5 redirect stubs + legacy directories from the Quanta→Build rename;
+ a stale INDEX.json pointing at a dead branch; publications.html was
+ nav-orphaned per the fix list (now in the sitemap — verify nav before citing).
+- The strongest traction number (Elder ENB: 900k+ downloads, 150k+ unique
+ users) is buried in a cover letter.
+
+### Asset verdict
+A world-class credibility engine with the commercial layer deliberately (or
+by omission) absent. Its value today is evidentiary: it wins the "are these
+people real" check instantly. Its unrealized value is everything after that
+check — measurement, capture, and a priced next step.
+
+## Property 2 — the Flywheel product surface
+
+### What it is
+The public `flywheel` engine repo, the `flywheel-desktop` client (v0.2.2,
+releases + Inno Setup installer), and the site's flywheel.html distribution
+page.
+
+### Strengths (VERIFIED)
+- A real, downloadable product: Windows x64 installer, 20 MB, published
+ SHA-256, engine frozen inside (no Python, no PATH). The only one-click
+ artifact in the portfolio.
+- Supply-chain honesty: releases built from version tags by a public pipeline
+ with a gate that refuses mislabeled artifacts; the unsigned-binary caveat is
+ published with the hash as the stated substitute.
+- OpenAI-compatible gateway (point an existing client at it; receipts arrive
+ under the same API) — the lowest-friction adoption path in the system.
+- Desktop client codebase enforces its own canon in CI (design tokens, verdict
+ mapping, and version truth are unit-tested; files under 300 lines).
+
+### Gaps (DRIFT)
+- **The public engine lags the dev engine by ~9×**: 11 API routes and v0.1.0
+ in the public repo vs. 96 routes in the local-model dev checkout. The
+ desktop client's README describes surfaces (receipts, plugins, memory,
+ studio, attest) that the public engine repo does not yet expose — an
+ integrity risk for exactly the audience this portfolio courts.
+- Distribution is GitHub-releases-only; no product page exists outside the
+ portfolio site; Windows-only installer today (macOS/Linux runners exist in
+ CI but no shipped artifact).
+- No pricing, license clarity beyond FSL-1.1-MIT on the engine, or support
+ channel.
+
+### Asset verdict
+The product is more real than its public face. The single cheapest
+credibility win in the portfolio is a dev→public sync cadence; the second is
+a product page with the installer, the hash, and one recorded demo above the
+fold.
+
+## Adjacent properties on the shelf
+
+Private repos `harpercompliance-site` and `sendmyletter-site` (both touched
+2026-07-15) indicate property experiments beyond the portfolio — the compliance
+name in particular aligns with vertical B and the owner's GRC documentation
+history. Not assessed (private, out of session scope); noted as existing
+assets when a domain strategy is chosen.
+
+## Recommendations (in order of compounding)
+
+1. **One name, one domain.** Pick the commercial identity (ZentropyLabs is the
+ publisher of record in the installer) and register one domain; serve the
+ site there with github.io as a mirror. Every DOI, receipt, and release
+ currently builds authority for a domain the project doesn't own.
+2. **Instrument honestly.** Self-hosted, privacy-respecting analytics plus one
+ conversion instrument (work-thread form or calendar). This is the site's
+ own credo applied to itself: no claim without measurement.
+3. **Publish the two vertical pages.** Vertical B assembles from seven
+ existing pages; vertical A (sales & marketing) does not exist and is
+ greenfield. The deck's Act II is the draft.
+4. **Sync the public engine.** Keep the public flywheel repo within one minor
+ version of the dev gateway so the desktop README's claims are checkable
+ against the code behind them.
+5. **Price one thing.** The security/ORCA practice is already revenue-shaped;
+ give it (or a receipts pilot for GRC/legal) a package and a number. One
+ priced offer converts the evidence library from portfolio into pipeline.
+6. **Promote the traction line.** Move the 900k-download graphics record and
+ the six-package PyPI roster onto the entry surfaces where a first-time
+ visitor forms their judgment.
diff --git a/docs/deck/flywheel-telos-deck.html b/docs/deck/flywheel-telos-deck.html
new file mode 100644
index 0000000..3216bf9
--- /dev/null
+++ b/docs/deck/flywheel-telos-deck.html
@@ -0,0 +1,599 @@
+Flywheel · Project Telos — The Whole Workflow
+
+
+
+
+
+
+
+
+
+
ZentropyLabs · Project Telos · 2026
+
Flywheel: the whole workflow, every tool its own product.
+
One local platform that routes any model, verifies what it can,
+ and hands back a receipt a stranger can re-run — presented here tool by tool, then as the system.
+
PREPARED 2026-07-30 · EVERY COUNT TRACEABLE TO A REPO OR PAGE · ← → to move
+
+
+
+
+
02 · The frontier problem
+
"Machines made claiming nearly free. Checking stayed expensive."
+
+
+
Generative systems now produce plans, code, summaries, campaigns, and citations faster than
+ any human review lane can absorb. Every industry is accumulating review debt: output
+ accepted on confidence, not on evidence.
+
The frontier question has flipped. It is no longer "can the model do it?" —
+ it is "who checks, against what, and can anyone re-check later?"
+
"Ask for the witness, not the confidence."
+
+
+
The verdict vocabulary
+
MATCH re-derived, byte for byte
+
DRIFT the record moved — caution, honestly shown
+
UNVERIFIABLE cannot be checked — never rounded up to "fine"
+
Never TRUSTED, by construction. The honest null is a first-class result.
03 · The practice — range first, rigor as the floor
+
Map the system → make the surface → test the claim → leave a usable artifact.
+
+
GitHub repositories
159
104 original · 55 forks · 37 private
+
Public engines
14
catalog.html, five-tier taxonomy
+
PyPI packages
6
the honest figure — self-corrected in public
+
DOI'd preprints
6
Zenodo · arXiv endorsement pending
+
Consumer track record
900k+
Elder ENB downloads · 150k+ unique users
+
Compiler proof point
1,521
passing tests, buildlang v1.2.0
+
+
One practice across model workflows, compilers, color science, real-time
+ graphics, research infrastructure, and clear writing — verification is one strong lane inside it, not the cage around it.
Each run leaves the next a better map, corpus, plan, verdict, route, or memory.
+
+
perceptiongather
→
+
verificationcrucible
→
+
orchestrationforum
→
+
structureindex
→
+
memorymneme · learn
→
+
creationstudio · telos
→
+
routing + surfaceflywheel · desktop
+ ⟲
+
+
Intake carries a provenance receipt. Judgment is a pure function that fails closed. Orchestration writes a
+ replayable causal ledger. The map carries file-and-line evidence. Memory carries its own provenance.
+ Creation is seeded and reproducible. Routing shows observed outcomes, not promises.
+
The loop is closed: flywheel loop-status → loop closure 9/9 (100%).
+ The model is the replaceable half; the loop is the durable one.
+
SOURCE · flywheel README · SUPERPROJECT.md (spine verified by test_superproject.py)
+
+
+
+
06 · Each tool alone — the working spine
+
+
Engine
In its own words
Standing
+
flywheel
"A companion for every model." Routes local or hosted, verifies what it can, escalates only the hard part, hands back a re-checkable receipt.
PUBLIC · DESKTOP v0.2.2
+
gather
"Reach it anywhere. Record how you got it." Guarded APIs, JS-walled pages, scans, audio, papers — every block carries a source hash.
SHIPPED 1.6.1 · PyPI
+
crucible
Register a thesis, steelman it, measure it against a substrate, track drift, refine the weakest axis. Verdicts fail closed.
VERSIONED 1.2.0 · PyPI
+
forum
An agent team with routing, quality checks, and a replayable causal hash-chain ledger. Human approval gates.
SHIPPED 1.13.0 · PyPI
+
index
Repo-inventory + dependency graph for many-repo workspaces; every edge carries file-and-line evidence. Fully offline.
BETA 2.9.0 · PyPI
+
emet
Re-derives bytes; reports MATCH / DRIFT / UNVERIFIABLE — "never TRUSTED, by construction." Four independent language implementations.
SHIPPED 1.1.0 · PyPI
+
learn
Turns your own material into a witnessed course; every graded step writes a receipt.
"The oracle decides, the model proposes. No receipt, no accept." A trained 14B coder behind the verified accept path.
DEV ENGINE · BUNDLED LANE
+
+
SOURCE · catalog.html · lane_identity.dart (each line stays in the product's own words)
+
+
+
+
08 · The private line — public where safe
+
Six organs stay gated, and the site says so plainly.
+
+
Runtime · ORCA
"Run the workspace, keep the record. And never touch the target." Security-engagement workbench — metadata-only, cannot reach the thing it assesses.
v1.0.0 · 361 tests · private line
+
Checkpoint · Aleph
One checkpoint over docs, package contracts, MCP declarations, CI, supply-chain evidence, receipts.
private line · public overview only
+
Boundary · behavior-transform
Mode-aware read/write/exec/fetch/input receipts. Raw content stays in local adapters; hosts receive counts, hashes, decisions, redacted refs.
public repo
+
Vault · Kun · Lab · Seed · Ledger · Sofer
Path-only receipts and rotation discipline; a controlled-lab assessment engine; an agent workflow ledger. Honestly labeled: awaiting a public-safe split.
private / proprietary
+
+
SOURCE · catalog.html organ index · orca.html · aleph.html
+
+
+
+
09 · Flywheel Desktop — one window over the platform
+
+
+
"No browser, no terminal."
+
A native desktop surface with 24 destinations — Chat, Compare, Code, Plan, Workflows,
+ Studio, Memory, Graph, Receipts, Lanes, and more — grouped by what you came to do:
+ START · DO · KNOW · ADVANCED.
+
+
Under every chat turn: the re-derivable receipt — request hash, prompt hash, response hash, served-model disclosure.
+
Tap any hash anywhere → jump to Receipts → Merkle inclusion proof, re-walkable offline.
+
Ships as one installer with the engine frozen inside — no Python, no PATH setup.
+
+
"The trust is built in, never in the way."
+
+
+
The accountability posture
+
+
Local-first: every call goes to 127.0.0.1:8799. Zero telemetry in the client.
+
Credentials: presence only, never values — typed once, stored in the OS keychain, never displayed, logged, or echoed back.
+
Capability gating: write and exec are grants, never defaults.
+
Honest nulls stay visible: "a view that cannot show DRIFT is not shipped."
10 · The connection surface — bigger than its roster
+
Seven lanes is the floor, not the ceiling.
+
+
Provisioned lanes
7 · health-probed via live MCP handshake
+
Mission flagships
11 · 5 live (doctor MATCH) + 6 declared
+
Public engines
14 · site catalog, 2026-07-29
+
Built-in connectors
17 MCP servers · filesystem → postgres → slack
+
Gateway routes
96 API routes in the dev engine
+
User plugins
unbounded · any MCP server registers as a plugin
+
+
Each count keeps its tier label — live, declared, built-in — because a roster
+ that rounds "declared" up to "live" would fail its own discipline. A down lane reports missing honestly; it never breaks the roster.
WE LEAD speed/cost THEY LEAD features · UX · traction
Benchmark-claim escrow: sealed verdict packets anyone can recompute offline — no incumbent escrows the claim itself.
+
telos agent end-to-end agents
Claude Code, OpenCode, Devin, OpenHands
WE LEAD native computer control THEY LEAD features · traction
The concurrent-desktop agent: a second pair of hands on your logged-in workstation while your keyboard stays free — every action a replayable evidence packet.
+
flywheel routers & runners
every router, every runner
"Routers route. Runners run. None of them check the work." The category's missing verb is verify — that is the whole wedge.
+
+
The battle map publishes THEY-LEAD rows on its own products. A tool that refuses to overclaim is a tool whose other claims you can trust.
Two markets are named below, and inside them the retired language comes back
+ on duty — because in these rooms it is not branding, it is the specification.
+
+ NO RECEIPT, NO ACCEPT
+ MODELS PROPOSE, ORACLES DISPOSE
+ MATCH · DRIFT · UNVERIFIABLE
+
+
"You can't grade your own homework. That's the anti-hallucination heart."
The industry that ships the most AI output owns the least proof of any of it.
+
+
Claims & brand safety
Every public sentence traceable to a frozen source: gather receipts + citation snapshots that outlive the live web. Substantiation for ad claims, on demand.
+
Agency accountability
forum runs campaign agent-work on a replayable causal ledger with human approval gates — the client re-checks the run instead of trusting the report.
+
AI-spend governance
flywheel routes to the cheapest honest source and keeps escalation receipts — the first per-provider scoreboard of observed cost and success, not promises.
+
Reproducible brand systems
studio-engine + telos: seeded generative campaigns, parametric typefaces, brand kits — same seed, same artifact, forever re-derivable.
+
Witnessed enablement
learn turns playbooks into graded courses where every completion writes a receipt — onboarding proof instead of completion checkboxes.
+
Account memory
mneme keeps client knowledge with provenance per memory — "who told us this, when, from what document" survives team turnover.
+
+
HONEST NOTE this vertical is greenfield: the current site
+ has zero sales/marketing surface. That is a gap and an opening — no positioning debt to unwind.
+
SOURCE · engine pages mapped against agency/martech workflows · site grep (no existing surface)
+
+
+
+
14 · Vertical B · Provenance-critical, privacy-bound industries
+
Where "who touched this, and can we prove it" is the workflow.
+
+
Sector
The live question
The system's answer
+
Compliance & model risk
Regulators require model origin and lineage to be auditable — incumbent platforms store lineage as editable database rows.
Merkle-logged receipts, offline inclusion proofs, hash-chained store with an audit tail. The record cannot quietly move.
+
Legal & e-discovery
"The summary is not the record."
Citation freezing, content-addressed corpora, attestation that binds sign-off to exactly what was walked — overclaims tracked explicitly.
+
Healthcare admin
Can a summary keep its uncertainty attached?
gather + forum lanes with honest-null rendering; UNVERIFIABLE stays visible instead of rounding up.
+
Newsrooms
Can every public sentence find its source?
gather → forum → crucible chain; every block carries a source hash; tamper is caught on re-read.
+
Security & red teams
Authorized offensive work needs records that hold up afterward.
ORCA engagement workbench (metadata-only, never touches the target) + a lawful-basis gate published in the open.
+
Regulated AI ops
Agents act faster than review; boundaries must be provable.
Boundary receipts (raw content never leaves local adapters), capability grants, presence-only credentials, zero telemetry, fully local loop.
The only base URL in the desktop client is 127.0.0.1:8799. No telemetry, no analytics, no outbound endpoint. Local model tiers are first-class; escalation is a choice with a receipt.
+
Keys are never collected
"Presence only, never values." A credential is typed once into an obscured field, handed to the OS keychain, and shown forever after as presence + source. The gateway refuses to accept values through the marketplace.
+
Proof works offline
A receipt is a Merkle leaf; the inclusion proof returns root + audit path so a third party recomputes it with no network, no account, no vendor.
+
Authority stays with the human
Write and exec are grants per run. Attestation binds sign-off to coverage. Comprehension gates mean you own what you can review, explain, and defend — pasting the diff back cannot pass.
+
+
"Built to be used humanely — and architected to resist being used any other way."
A real product whose public face runs one lap behind the engine.
+
+
+
+
A downloadable installer (Windows x64, 20 MB) with a published SHA-256 and an honest unsigned-binary note — "the receipt does the signature's job."
+
Desktop v0.2.2 releases built from version tags by a public pipeline that refuses to ship a mislabeled artifact.
+
OpenAI-compatible gateway: point any existing client at it and get receipts under the same API.
+
+
+
The public flywheel repo exposes 11 API routes; the dev engine has 96. The public surface undersells the shipped product by roughly a factor of nine.
+
Distribution is GitHub-releases-only; no product page outside the portfolio site.
+
+
+
+
Route parity, public vs dev
+
+
public repo
11 routes
+
dev engine
96 routes
+
+
A sync cadence from dev to public is the single cheapest credibility win available.
+
+
+
SOURCE · flywheel.html · gateway.py route grep, both checkouts, 2026-07-30
+
+
+
+
20 · Recommendations — in verdict order
+
+
#
Move
Why it compounds
+
01
One name, one domain. Consolidate Harper / Project Telos / ZentropyLabs under a single registered domain; keep github.io as a mirror.
Every paper, receipt, and release currently builds authority for a domain you don't own.
+
02
Instrument honestly. Privacy-respecting, self-hosted analytics + one conversion instrument (a real work-thread form or calendar).
The portfolio's own thesis: no claim without measurement. The site cannot currently make a measured claim about itself.
+
03
Publish the two vertical pages. Assemble the scattered provenance material into one buyer-facing argument; write the first sales-&-marketing surface (it does not exist yet).
Vertical B's evidence is already written — it lives on seven different pages. Vertical A is an open field.
+
04
Sync the public engine. A release cadence that keeps the public flywheel repo within one minor version of the dev gateway.
Closes the 11-vs-96 gap; the README's claims become checkable against the code behind them.
+
05
Name a price for one thing. The security/ORCA practice is already revenue-shaped; give it (or a receipts pilot) a package and a number.
A single priced offer converts the whole evidence library from portfolio into pipeline.
+
+
Also on the shelf, unexplored: two private site repos (harpercompliance-site, sendmyletter-site) suggest adjacent property experiments already underway.
+
SOURCE · property assessment 2026-07-30
+
+
+
+
21 · Honest boundaries — what this deck does not claim
+
+
Published nulls stay published
Flywheel's own uplift test: +0.100 with a 95% interval of [−0.236, +0.420] — includes zero, so no capability uplift is claimed. The site prints this against its own product.
+
Maturity labels held
Live ≠ declared ≠ forming. The installer is not code-signed (the hash is the substitute). arXiv listings are endorsement-pending. Demos are recorded workflows, not scale benchmarks.
+
Traction is early
Six PyPI packages, one downloadable product, zero pricing to date. The 900k-download record is from the graphics line, not the platform.
+
A deliberate register divergence
BRAND-NARRATIVE.md retires verification-first language at site level. Act II of this deck re-arms it on purpose, as vertical sales language — flagged here so the divergence is a decision, not a drift.
+
+
From 4fea5d28dfd690fc24f2febde55c61cd7e738123 Mon Sep 17 00:00:00 2001
From: Claude
Date: Thu, 30 Jul 2026 03:18:29 +0000
Subject: [PATCH 2/3] docs: reskin deck to the current ZentropyLabs site canon
The first cut styled the deck from the site's superseded 2026-07-09 dark
cascade (violet ground, lime primary). This reskin matches the live
2026-07-19 zentropy-sitewide layer: #070406 ground with teal/rust blooms,
cyan #8ee3f2 brand accent with lime reserved for verdicts, Hanken Grotesk +
Conso + ZentropyDisplay embedded as data URIs, the notched oxblood brand
plate with the real aperture mark as the cover device, fade-out gradient
hairlines, and the site's motion grammar (reveal, cardblob, specular rims)
with reduced-motion clamps. Slide content unchanged.
Co-Authored-By: Claude Fable 5
Claude-Session: https://claude.ai/code/session_01PbFKzRRinxkZ34EwoA8Jwf
---
docs/deck/flywheel-telos-deck.html | 387 ++++++++++++++++-------------
1 file changed, 219 insertions(+), 168 deletions(-)
diff --git a/docs/deck/flywheel-telos-deck.html b/docs/deck/flywheel-telos-deck.html
index 3216bf9..9a8408c 100644
--- a/docs/deck/flywheel-telos-deck.html
+++ b/docs/deck/flywheel-telos-deck.html
@@ -1,159 +1,227 @@
Flywheel · Project Telos — The Whole Workflow
@@ -161,13 +229,21 @@
-
-
-
ZentropyLabs · Project Telos · 2026
-
Flywheel: the whole workflow, every tool its own product.
-
One local platform that routes any model, verifies what it can,
- and hands back a receipt a stranger can re-run — presented here tool by tool, then as the system.
-
PREPARED 2026-07-30 · EVERY COUNT TRACEABLE TO A REPO OR PAGE · ← → to move
+
+
+
zentropyLabsProject Telos · 2026
+
Flywheel: the whole workflow, every tool its own product.
+
One local platform that routes any model, verifies what it can,
+ and hands back a receipt a stranger can re-run — presented tool by tool, then as the system.
+
Prepared 2026-07-30 · every count traceable · ← → to move
+
+
+
+
+
+ Zentropy mark · approved static artwork
+
+
@@ -184,14 +260,14 @@
Flywheel: the whole workflow, every tool its own product.
"Ask for the witness, not the confidence."
-
The verdict vocabulary
+
The verdict vocabulary
MATCH re-derived, byte for byte
DRIFT the record moved — caution, honestly shown
UNVERIFIABLE cannot be checked — never rounded up to "fine"
Never TRUSTED, by construction. The honest null is a first-class result.
Map the system → make the surface → test the claim → leave a usable ar
One practice across model workflows, compilers, color science, real-time
graphics, research infrastructure, and clear writing — verification is one strong lane inside it, not the cage around it.
Each run leaves the next a better map, corpus, plan, verdict, route, or memo
Creation is seeded and reproducible. Routing shows observed outcomes, not promises.
The loop is closed: flywheel loop-status → loop closure 9/9 (100%).
The model is the replaceable half; the loop is the durable one.
-
SOURCE · flywheel README · SUPERPROJECT.md (spine verified by test_superproject.py)
+
Source · flywheel README · SUPERPROJECT.md (spine verified by test_superproject.py)
06 · Each tool alone — the working spine
-
+
Engine
In its own words
Standing
flywheel
"A companion for every model." Routes local or hosted, verifies what it can, escalates only the hard part, hands back a re-checkable receipt.
PUBLIC · DESKTOP v0.2.2
gather
"Reach it anywhere. Record how you got it." Guarded APIs, JS-walled pages, scans, audio, papers — every block carries a source hash.
SHIPPED 1.6.1 · PyPI
@@ -258,12 +334,12 @@
Each run leaves the next a better map, corpus, plan, verdict, route, or memo
emet
Re-derives bytes; reports MATCH / DRIFT / UNVERIFIABLE — "never TRUSTED, by construction." Four independent language implementations.
SHIPPED 1.1.0 · PyPI
learn
Turns your own material into a witnessed course; every graded step writes a receipt.
"The oracle decides, the model proposes. No receipt, no accept." A trained 14B coder behind the verified accept path.
DEV ENGINE · BUNDLED LANE
-
SOURCE · catalog.html · lane_identity.dart (each line stays in the product's own words)
+
Source · catalog.html · lane_identity.dart (each line stays in the product's own words)
@@ -286,7 +362,7 @@
Six organs stay gated, and the site says so plainly.
Boundary · behavior-transform
Mode-aware read/write/exec/fetch/input receipts. Raw content stays in local adapters; hosts receive counts, hashes, decisions, redacted refs.
public repo
Vault · Kun · Lab · Seed · Ledger · Sofer
Path-only receipts and rotation discipline; a controlled-lab assessment engine; an agent workflow ledger. Honestly labeled: awaiting a public-safe split.
private / proprietary
-
SOURCE · catalog.html organ index · orca.html · aleph.html
+
Source · catalog.html organ index · orca.html · aleph.html
@@ -305,7 +381,7 @@
"No browser, no terminal."
"The trust is built in, never in the way."
-
The accountability posture
+
The accountability posture
Local-first: every call goes to 127.0.0.1:8799. Zero telemetry in the client.
Credentials: presence only, never values — typed once, stored in the OS keychain, never displayed, logged, or echoed back.
Each count keeps its tier label — live, declared, built-in — because a roster
that rounds "declared" up to "live" would fail its own discipline. A down lane reports missing honestly; it never breaks the roster.
Two markets are named below, and inside them the retired language comes back
on duty — because in these rooms it is not branding, it is the specification.
@@ -359,11 +435,11 @@
Here the portfolio becomes an offer.
MATCH · DRIFT · UNVERIFIABLE
"You can't grade your own homework. That's the anti-hallucination heart."
The industry that ships the most AI output owns the least proof of any of it.
Claims & brand safety
Every public sentence traceable to a frozen source: gather receipts + citation snapshots that outlive the live web. Substantiation for ad claims, on demand.
@@ -375,11 +451,11 @@
The industry that ships the most AI output owns the least proof of any of it
HONEST NOTE this vertical is greenfield: the current site
has zero sales/marketing surface. That is a gap and an opening — no positioning debt to unwind.
-
SOURCE · engine pages mapped against agency/martech workflows · site grep (no existing surface)
+
Source · engine pages mapped against agency/martech workflows · site grep (no existing surface)
-
14 · Vertical B · Provenance-critical, privacy-bound industries
+
14 · Vertical B · Provenance-critical, privacy-bound industries
Where "who touched this, and can we prove it" is the workflow.
Sector
The live question
The system's answer
@@ -390,11 +466,11 @@
Where "who touched this, and can we prove it" is the workflow.
Security & red teams
Authorized offensive work needs records that hold up afterward.
ORCA engagement workbench (metadata-only, never touches the target) + a lawful-basis gate published in the open.
Regulated AI ops
Agents act faster than review; boundaries must be provable.
Boundary receipts (raw content never leaves local adapters), capability grants, presence-only credentials, zero telemetry, fully local loop.
The only base URL in the desktop client is 127.0.0.1:8799. No telemetry, no analytics, no outbound endpoint. Local model tiers are first-class; escalation is a choice with a receipt.
Keys are never collected
"Presence only, never values." A credential is typed once into an obscured field, handed to the OS keychain, and shown forever after as presence + source. The gateway refuses to accept values through the marketplace.
@@ -402,11 +478,11 @@
Where "who touched this, and can we prove it" is the workflow.
Authority stays with the human
Write and exec are grants per run. Attestation binds sign-off to coverage. Comprehension gates mean you own what you can review, explain, and defend — pasting the diff back cannot pass.
"Built to be used humanely — and architected to resist being used any other way."
A real product whose public face runs one lap behind the engine.
-
Route parity, public vs dev
+
Route parity, public vs dev
public repo
11 routes
dev engine
96 routes
@@ -486,7 +562,7 @@
A real product whose public face runs one lap behind the engine.
A sync cadence from dev to public is the single cheapest credibility win available.
-
SOURCE · flywheel.html · gateway.py route grep, both checkouts, 2026-07-30
+
Source · flywheel.html · gateway.py route grep, both checkouts, 2026-07-30
@@ -500,7 +576,7 @@
A real product whose public face runs one lap behind the engine.
05
Name a price for one thing. The security/ORCA practice is already revenue-shaped; give it (or a receipts pilot) a package and a number.
A single priced offer converts the whole evidence library from portfolio into pipeline.
Also on the shelf, unexplored: two private site repos (harpercompliance-site, sendmyletter-site) suggest adjacent property experiments already underway.
-
SOURCE · property assessment 2026-07-30
+
Source · property assessment 2026-07-30
@@ -511,12 +587,12 @@
A real product whose public face runs one lap behind the engine.
Traction is early
Six PyPI packages, one downloadable product, zero pricing to date. The 900k-download record is from the graphics line, not the platform.
A deliberate register divergence
BRAND-NARRATIVE.md retires verification-first language at site level. Act II of this deck re-arms it on purpose, as vertical sales language — flagged here so the divergence is a decision, not a drift.
else if (e.key === 'Home') { show(0); }
else if (e.key === 'End') { show(n - 1); }
});
- // click-to-advance on the right half of a slide, back on the far-left edge
document.getElementById('deck').addEventListener('click', function(e){
if (e.target.closest('a, button, .tblwrap')) return;
var x = e.clientX / window.innerWidth;
if (x > 0.66) show(i + 1); else if (x < 0.08) show(i - 1);
});
show(0);
-
- /* Cover field: the aperture's own generator — mulberry32, seed 58.
- A dithered corona of ink dots around a void core, one verified flare arc.
- Deterministic: same seed, same mark. Spectrum lives only here (site rule). */
- function mulberry32(a){ return function(){ a|=0; a=a+0x6D2B79F5|0;
- var t=Math.imul(a^a>>>15,1|a); t=t+Math.imul(t^t>>>7,61|t)^t;
- return ((t^t>>>14)>>>0)/4294967296; } }
- var cv = document.getElementById('cover-field');
- if (cv && cv.getContext){
- var ctx = cv.getContext('2d'), R = mulberry32(58), W = cv.width, C = W/2;
- for (var d = 0; d < 460; d++){
- var ang = R()*Math.PI*2;
- var rad = C*0.36 + Math.pow(R(),1.6)*C*0.5;
- var sz = 0.6 + R()*2.1;
- var hueRoll = R();
- ctx.fillStyle = hueRoll < 0.06 ? 'oklch(87% .215 133/.85)'
- : hueRoll < 0.10 ? 'oklch(76% .19 46/.6)'
- : 'oklch(90% .02 318/' + (0.12 + R()*0.5) + ')';
- ctx.beginPath();
- ctx.arc(C + Math.cos(ang)*rad, C + Math.sin(ang)*rad, sz, 0, 7);
- ctx.fill();
- }
- ctx.strokeStyle = 'oklch(87% .215 133/.9)'; ctx.lineWidth = 2.5;
- ctx.beginPath(); ctx.arc(C, C, C*0.37, -0.9, -0.15); ctx.stroke();
- ctx.strokeStyle = 'oklch(62% .08 308/.35)'; ctx.lineWidth = 1;
- ctx.beginPath(); ctx.arc(C, C, C*0.34, 0, 7); ctx.stroke();
- }
})();
From a3a45a71e027f4d4e93803ba5c3ed3cb8ee06252 Mon Sep 17 00:00:00 2001
From: Zain Dana Harper <17142659+HarperZ9@users.noreply.github.com>
Date: Wed, 29 Jul 2026 21:04:37 -0700
Subject: [PATCH 3/3] docs: ground the deck in verified local state, taking
over the cloud draft
Takeover of the claude.ai/code session's deck (branch
claude/flywheel-deck-domain-analysis-9lpy1a), which was authored well but
isolated from local ground truth. Corrections, each verified locally:
- buildlang test count 1,521 -> 1,605 (the cloud draft inherited the
stale figure from buildlang.html, itself fixed today in the site's PR
90; both surfaces now quote main's own stated baseline)
- dev gateway route count 96 -> 102 (measured: unique route-path
literals in the dev gateway's dispatch, local census 2026-07-30; the
source line now names the real measurement instead of a grep the
cloud session could not have run)
- slide 5 gains the live proof the isolation hid: the 2026-07-29 round
trip where a local model proposed through a sealed boundary, a
model-free oracled kernel disposed, and one receipt chain bound both
- slide 7: buildlang row carries the five-modes sealed-receipt chain;
local-model row carries the model-boundary receipt, labeled in review
- register sweep: 111 em-dashes removed across the deck and the four
analyses, per the public-surface canon; no other prose changed
The cloud draft's structure, three-act narrative, analyses, and honest
verdicts (THEY-LEAD rows, published nulls, greenfield admissions) are
kept intact; they held up under audit.
Co-Authored-By: Claude Fable 5
---
docs/deck/01-github-profile-analysis.md | 36 +++----
docs/deck/02-vertical-value.md | 22 ++--
docs/deck/03-target-audiences.md | 20 ++--
docs/deck/04-property-assessment.md | 20 ++--
docs/deck/flywheel-telos-deck.html | 138 ++++++++++++------------
5 files changed, 118 insertions(+), 118 deletions(-)
diff --git a/docs/deck/01-github-profile-analysis.md b/docs/deck/01-github-profile-analysis.md
index cd8cbb5..2d601fe 100644
--- a/docs/deck/01-github-profile-analysis.md
+++ b/docs/deck/01-github-profile-analysis.md
@@ -1,4 +1,4 @@
-# GitHub profile analysis — HarperZ9 / ZentropyLabs
+# GitHub profile analysis -- HarperZ9 / ZentropyLabs
> Census taken 2026-07-30 from the live account listing. Counts are exact at
> that timestamp. Companion files: 02-vertical-value.md, 03-target-audiences.md,
@@ -16,28 +16,28 @@
## The five-tier taxonomy (per the site's own catalog)
-### Tier 1 — the fourteen public engines
+### Tier 1 -- the fourteen public engines
flywheel (route + verify; desktop client v0.2.2), telos (the shared workbench,
live v0.2.0), index (2.9.0 beta, PyPI `index-graph`), gather (1.6.1 shipped,
PyPI `gather-engine`), forum (1.13.0 shipped, PyPI `forum-engine`), crucible
(1.2.0, PyPI `crucible-bench`), emet (1.1.0, PyPI `emet`, four independent
-language implementations), buildlang (v1.2.0, 1,521 passing tests), learn
+language implementations), buildlang (v1.2.0, 1,605 passing tests), learn
(v1.6.0), relay, plexus, mneme, studio-engine, build-color (1.0.2 beta,
PyPI `build-color`).
-Six confirmed PyPI packages — the site publicly corrected an earlier "19"
+Six confirmed PyPI packages -- the site publicly corrected an earlier "19"
over-count to this honest figure.
-### Tier 2 — the private line ("public where safe")
-- **orca** (Runtime) — security-engagement workbench, v1.0.0, 361 tests;
+### Tier 2 -- the private line ("public where safe")
+- **orca** (Runtime) -- security-engagement workbench, v1.0.0, 361 tests;
metadata-only, cannot reach the target it assesses. The most revenue-shaped
private asset.
-- **aleph** (Checkpoint) — release/supply-chain coordination across docs,
+- **aleph** (Checkpoint) -- release/supply-chain coordination across docs,
contracts, MCP declarations, CI, receipts.
-- **kun** (Vault) — path-only receipts, rotation discipline, value-free.
-- **behavior-transform.io** (Boundary) — public repo; mode-aware
+- **kun** (Vault) -- path-only receipts, rotation discipline, value-free.
+- **behavior-transform.io** (Boundary) -- public repo; mode-aware
read/write/exec/fetch/input receipts; raw content stays in local adapters.
-- **seed** (Lab) and **sofer** (Ledger) — private/proprietary; honestly labeled
+- **seed** (Lab) and **sofer** (Ledger) -- private/proprietary; honestly labeled
as needing a public-safe split (credential-pattern corpora inside).
- Adjacent private security cluster: offensive-platform, red-team-platform,
adversarial, protected-corridors, protected-research, legal-intl,
@@ -46,23 +46,23 @@ over-count to this honest figure.
ai-safety-prefire, ai-safety-guardrail-manager, unified-engine, aurora,
apps, flywheel-host.
-### Tier 3 — substrates
-proof-surface (652 tests, zero deps — the shared proof-packet foundation),
+### Tier 3 -- substrates
+proof-surface (652 tests, zero deps -- the shared proof-packet foundation),
coherence-membrane, accountable-surface, accountable-engine (Forming; "the
inward half is owed"), provenance-sensorium, reconcile, raw (D3D11 mid-frame
graphics proxy), witnessing-spine (five regulated-sector adversarial
steelmans, DOI 10.5281/zenodo.20778927), senses-and-sensibility,
faithful-transpile, coherence and kernels lines (signal-kernels,
-anomaly-kernels — C++23 header-only).
+anomaly-kernels -- C++23 header-only).
-### Tier 4 — bricks (release + agent toolkit)
+### Tier 4 -- bricks (release + agent toolkit)
secret-redact-io, release-surface-scanner, public-surface-sweeper,
repo-proof-index, proof-surface-report, model-provenance-validator,
gpu-trace-validator, agent-audit, agent-routing-kit, agent-hook-pack,
context-curator-lite, workflow-harness-lite, consulting-template-kit,
calibrate-pro, wol-pi.
-### Tier 5 — research + receipts-in-the-wild
+### Tier 5 -- research + receipts-in-the-wild
Receipt repos aimed at other ecosystems: crewai-kickoff-replay-receipt,
vllm-ci-forum-receipt. Research surface lives on the site: 3 theses,
10 experiment packets, 6 DOI'd preprints (arXiv endorsement-pending),
@@ -91,7 +91,7 @@ scorekeeper), MCP ecosystem curation (a dozen awesome-mcp-* lists, registry,
python-sdk), core Python tooling (numpy, poetry, setuptools, isort, tomlkit,
typeguard, grimp, datasette, llm), and agent-security reading
(system_prompts_leaks, awesome-ai-agents-security). Read as a map: the
-account tracks exactly the frontier its products verify — agents, evals,
+account tracks exactly the frontier its products verify -- agents, evals,
MCP, and the supply chain under them.
## Engine-state cross-check (dev vs public)
@@ -111,8 +111,8 @@ DRIFT); 85 + 243 passing tests on that slice; CI across three OSes.
## Reading the whole profile in one line
-One person, one discipline — perceive, transform under a named criterion,
-carry a re-checkable receipt — applied across model infrastructure, compilers,
+One person, one discipline -- perceive, transform under a named criterion,
+carry a re-checkable receipt -- applied across model infrastructure, compilers,
color science, graphics, security practice, and research, with the honesty
mechanics (published nulls, self-caught over-counts, maturity tiers) intact
everywhere. The portfolio's weakness is not depth or range; it is that no
diff --git a/docs/deck/02-vertical-value.md b/docs/deck/02-vertical-value.md
index 0de40a2..54fc380 100644
--- a/docs/deck/02-vertical-value.md
+++ b/docs/deck/02-vertical-value.md
@@ -1,4 +1,4 @@
-# Vertical value assessment — two industries, each tool alone and as the system
+# Vertical value assessment -- two industries, each tool alone and as the system
> Companion to 01-github-profile-analysis.md. The claim discipline of the
> portfolio applies to this document: where a value statement is a projection
@@ -8,16 +8,16 @@
Machines made claiming nearly free; checking stayed expensive. Both verticals
below are drowning in AI-produced output whose acceptance currently rests on
-confidence. The system's one operation — propose, verify against an external
-criterion, keep a re-checkable receipt — is the missing accept path. The
+confidence. The system's one operation -- propose, verify against an external
+criterion, keep a re-checkable receipt -- is the missing accept path. The
pitch deck already names the wedge categories (originality-verification,
accountable-agent-action, provenance) and is honest that they have no analyst
TAM yet: that is the whitespace, and also the risk.
-## Vertical A — sales & marketing (greenfield)
+## Vertical A -- sales & marketing (greenfield)
Grep-verified: the current site has zero sales/marketing surface area. No
-positioning debt, no incumbent framing to unwind — and no existing proof of
+positioning debt, no incumbent framing to unwind -- and no existing proof of
demand either. Everything below is a mapping of shipped capabilities onto
known agency/martech pains, not observed traction.
@@ -26,7 +26,7 @@ known agency/martech pains, not observed traction.
| **gather** + snapshot (citation freezing) | research intake with provenance receipts | Claims substantiation: every public sentence traceable to a frozen, hashed source that outlives the live web. Ad-claim defense files that assemble themselves. |
| **forum** | agent orchestration on a replayable causal hash-chain ledger with human gates | Agency accountability: campaign automation the client can re-check instead of trusting a report. A deliverable format no incumbent offers. |
| **flywheel** routing + receipts | one request shape over every provider, observed per-provider scoreboard | AI-spend governance: cost-honest routing, escalation receipts, the first "what did the model spend buy us" ledger. |
-| **studio-engine** + telos creative | seeded, replayable generative worlds/brand kits | Reproducible brand systems: same seed, same artifact — campaign creative that is re-derivable, auditable, and provably original-to-seed. |
+| **studio-engine** + telos creative | seeded, replayable generative worlds/brand kits | Reproducible brand systems: same seed, same artifact -- campaign creative that is re-derivable, auditable, and provably original-to-seed. |
| **learn** | witnessed courses; every graded step writes a receipt | Enablement proof: onboarding and certification that produce receipts, not completion checkboxes. |
| **mneme** | agent memory where every memory carries provenance | Account knowledge that survives turnover: "who told us this, when, from which document" as a queryable property. |
| **crucible** | falsifiable claim verification | Marketing-claims QA and competitor-claim testing: register the claim, measure it, publish the verdict packet. |
@@ -37,11 +37,11 @@ verification (crucible) → campaign orchestration (forum) → creative (studio)
record. The sale is not "AI tools"; it is *defensible delivery*.
**Honest boundaries**: no marketing-specific integrations exist today (no CRM,
-no ad-platform connectors — though 17 built-in MCP connectors plus the plugin
+no ad-platform connectors -- though 17 built-in MCP connectors plus the plugin
registry are the extension path). First proof would need one design-partner
agency.
-## Vertical B — provenance-critical, privacy-bound industries (assembled, not built)
+## Vertical B -- provenance-critical, privacy-bound industries (assembled, not built)
The material for this vertical already exists across seven site pages and
several repos; it has never been assembled into one buyer-facing argument.
@@ -55,7 +55,7 @@ several repos; it has never been assembled into one buyer-facing argument.
| **Security / red teams (gov, law enforcement, AI labs)** | Authorized adversarial work needs records that hold up afterward, and a lawful-basis gate. | The existing security.html practice + ORCA (v1.0.0, 361 tests, metadata-only engagement workbench). Already revenue-shaped. |
| **Regulated AI operations** | Agent boundaries must be provable, credentials must not travel. | Boundary receipts (raw content never leaves local adapters), capability grants per run, presence-only credentials in the OS keychain, zero telemetry, fully local loop. |
-**The privacy story is architectural, not contractual** — this is the
+**The privacy story is architectural, not contractual** -- this is the
differentiator dense-workflow industries actually buy:
- The desktop client's only base URL is 127.0.0.1:8799; no outbound endpoint
exists in the codebase.
@@ -66,14 +66,14 @@ differentiator dense-workflow industries actually buy:
- Write/exec are grants per run; a verify stage without an exec grant reports
UNVERIFIABLE rather than pretending.
-**As the system**: the same loop, pointed at regulated work — intake with
+**As the system**: the same loop, pointed at regulated work -- intake with
receipts, judgment that fails closed, orchestration with human gates, sign-off
bound to coverage, and an audit trail that a regulator, opposing counsel, or
incident reviewer can re-derive without trusting the operator.
## Value of each tool alone vs. the system
-Each lane is deliberately "a full product that also runs alone" — gather,
+Each lane is deliberately "a full product that also runs alone" -- gather,
crucible, index, forum, learn are individually pip-installable with their own
receipts. Alone, each competes in a crowded category on speed/cost and honesty
(the battle map publishes THEY_LEAD rows). Together they occupy compositions
diff --git a/docs/deck/03-target-audiences.md b/docs/deck/03-target-audiences.md
index d800a2f..c4d9b41 100644
--- a/docs/deck/03-target-audiences.md
+++ b/docs/deck/03-target-audiences.md
@@ -1,4 +1,4 @@
-# Target audiences — where, how, and why these tools matter on today's frontier
+# Target audiences -- where, how, and why these tools matter on today's frontier
> The frontier condition in one line: autonomous systems now produce work
> faster than any review lane absorbs it, provenance duties are hardening into
@@ -10,33 +10,33 @@
| Pressure (where the frontier bites) | The system's answer |
|---|---|
-| **Agents outran review** — autonomous work lands faster than humans check it | The verified accept path: an external oracle decides; models propose, oracles dispose; UNVERIFIABLE is a legal, visible outcome |
-| **Provenance became law** — lineage/origin duties moving into regulation across finance, health, media | Receipts as infrastructure: content-addressed, Merkle-proven, offline-verifiable, vendor-independent |
-| **Claiming became free** — capabilities, benchmarks, and originality asserted daily, checked never | Claim escrow: sealed verdict packets a stranger, competitor, or court can recompute |
+| **Agents outran review** -- autonomous work lands faster than humans check it | The verified accept path: an external oracle decides; models propose, oracles dispose; UNVERIFIABLE is a legal, visible outcome |
+| **Provenance became law** -- lineage/origin duties moving into regulation across finance, health, media | Receipts as infrastructure: content-addressed, Merkle-proven, offline-verifiable, vendor-independent |
+| **Claiming became free** -- capabilities, benchmarks, and originality asserted daily, checked never | Claim escrow: sealed verdict packets a stranger, competitor, or court can recompute |
## Audience matrix
| Who | Where they live | Why now | How they enter | Value to them | Value to us |
|---|---|---|---|---|---|
-| **Agency ops lead** | Campaign delivery, client reporting | Clients audit AI-made work; agencies have no proof layer | forum demo: the replayable run ledger | Defensible delivery — the client re-checks instead of trusts | First design partner in the greenfield vertical |
+| **Agency ops lead** | Campaign delivery, client reporting | Clients audit AI-made work; agencies have no proof layer | forum demo: the replayable run ledger | Defensible delivery -- the client re-checks instead of trusts | First design partner in the greenfield vertical |
| **CMO / martech owner** | Brand, content, AI-spend budgets | AI spend unmeasured; claims risk rising | flywheel routing scoreboard + gather claim receipts | Cost + substantiation in one surface | The paying wedge account for vertical A |
| **GRC / model-risk officer** | Banks, insurers, model inventories | Lineage regulation arrived; current audit trails are editable rows | Witnessing Spine findings + a receipts inclusion-proof demo | Tamper-evident lineage a regulator can re-derive | The regulated flagship reference |
| **Legal ops / e-discovery lead** | Firms, corporate legal departments | AI summaries entering the record without their sources | snapshot (citation freezing) + attestation | Summaries bound to frozen records; sign-off bound to coverage | The dense-workflow proving ground |
| **Newsroom standards editor** | Media, fact-check desks | Generated content floods intake; provenance IS the product | field-guide media lane: gather → forum → crucible | Source-checked publishing with per-block hashes | A public credibility case study |
| **Red team / security lead** | Government, law enforcement, AI labs | AI attack-surface work needs lawful, replayable records | ORCA + the security.html engagement gate | Engagement records that hold up afterward | The already-revenue-shaped practice |
| **Clinical admin / health-ops** | Hospital administration, payer ops | Summaries must keep uncertainty attached | field-guide clinical lane (gather + forum) | Honest-null summaries; UNVERIFIABLE stays visible | Entry into the highest-stakes provenance market |
-| **Builders & researchers** | OSS, eval/agent infrastructure | Already served — the audience the site speaks to today | PyPI packages, recorded demos, DOI'd papers | Tools + receipts + a falsifiable research program | Contributors, field testers, standing — not revenue |
+| **Builders & researchers** | OSS, eval/agent infrastructure | Already served -- the audience the site speaks to today | PyPI packages, recorded demos, DOI'd papers | Tools + receipts + a falsifiable research program | Contributors, field testers, standing -- not revenue |
## Where each audience is reached
- **Agencies / CMOs**: martech and agency-ops communities; the deck's Act II-A
- is the first artifact that speaks to them — nothing on the site does yet.
+ is the first artifact that speaks to them -- nothing on the site does yet.
- **GRC / model risk**: the Witnessing Spine paper is citable; model-risk and
AI-governance forums; the "editable rows vs. Merkle proof" demo is the hook.
- **Legal**: e-discovery and legal-ops conferences; the citation-freezing demo
is self-explanatory to anyone who has fought over a dead link in a record.
- **Newsrooms**: the field guide's media lane is already written as a script
- with a break test — publishable as-is.
+ with a break test -- publishable as-is.
- **Security**: the practice exists with a lawful-basis gate; the audience is
reached by referral and credential, not marketing.
- **Builders**: already arriving via PyPI, the demos, and the papers.
@@ -47,14 +47,14 @@ The audience that already responds (builders, researchers) is not the audience
that pays. The audiences that would pay (compliance, legal, agencies, regulated
ops) are named nowhere on the current public surface except behind the
security page's authorization gate. Closing that gap is a positioning task,
-not an engineering one — the evidence is already written, scattered across
+not an engineering one -- the evidence is already written, scattered across
seven pages.
## Why this matters to each side (the deck's dual promise)
For the owner: each tool is independently sellable (pip-installable, own
receipts, own demo), so every audience above is reachable without waiting for
-platform adoption. For the audience: the system never asks to be trusted —
+platform adoption. For the audience: the system never asks to be trusted --
every claim it makes about their work arrives with the means to re-check it,
which is precisely the property their regulators, clients, editors, and
courts are beginning to demand.
diff --git a/docs/deck/04-property-assessment.md b/docs/deck/04-property-assessment.md
index c6ab8aa..2ad21fc 100644
--- a/docs/deck/04-property-assessment.md
+++ b/docs/deck/04-property-assessment.md
@@ -1,4 +1,4 @@
-# Web-property assessment — the two domains
+# Web-property assessment -- the two domains
> "Both domains" per the owner's direction: (1) the harperz9.github.io
> portfolio/Project Telos site, and (2) the Flywheel product surface
@@ -6,7 +6,7 @@
> Assessed 2026-07-30 from full clones. No registered custom domain exists
> today; behavior-transform.io appears only as a GitHub repository name.
-## Property 1 — harperz9.github.io (Project Telos site)
+## Property 1 -- harperz9.github.io (Project Telos site)
### What it is
72 sitemap pages (87 HTML files), actively maintained (PR #89 merged
@@ -33,7 +33,7 @@ private telos-v2 source with a verification gate.
brand identity split three ways (Zain Dana Harper / Project Telos /
ZentropyLabs, with a ZentropyLabs-ai GitHub org besides).
- **Zero analytics** (grep-verified: no GA/GTM/Plausible/Umami/Fathom/Clarity).
- The site is unmeasurable — by its own thesis, it cannot make a measured
+ The site is unmeasurable -- by its own thesis, it cannot make a measured
claim about itself.
- **No conversion instrument.** No form, calendar, newsletter, or pricing;
every CTA terminates at a gmail address. The consulting CTA has no landing
@@ -42,7 +42,7 @@ private telos-v2 source with a verification gate.
count rather than a value proposition.
- Residue: 5 redirect stubs + legacy directories from the Quanta→Build rename;
a stale INDEX.json pointing at a dead branch; publications.html was
- nav-orphaned per the fix list (now in the sitemap — verify nav before citing).
+ nav-orphaned per the fix list (now in the sitemap -- verify nav before citing).
- The strongest traction number (Elder ENB: 900k+ downloads, 150k+ unique
users) is buried in a cover letter.
@@ -50,9 +50,9 @@ private telos-v2 source with a verification gate.
A world-class credibility engine with the commercial layer deliberately (or
by omission) absent. Its value today is evidentiary: it wins the "are these
people real" check instantly. Its unrealized value is everything after that
-check — measurement, capture, and a priced next step.
+check -- measurement, capture, and a priced next step.
-## Property 2 — the Flywheel product surface
+## Property 2 -- the Flywheel product surface
### What it is
The public `flywheel` engine repo, the `flywheel-desktop` client (v0.2.2,
@@ -67,15 +67,15 @@ page.
with a gate that refuses mislabeled artifacts; the unsigned-binary caveat is
published with the hash as the stated substitute.
- OpenAI-compatible gateway (point an existing client at it; receipts arrive
- under the same API) — the lowest-friction adoption path in the system.
+ under the same API) -- the lowest-friction adoption path in the system.
- Desktop client codebase enforces its own canon in CI (design tokens, verdict
mapping, and version truth are unit-tested; files under 300 lines).
### Gaps (DRIFT)
- **The public engine lags the dev engine by ~9×**: 11 API routes and v0.1.0
- in the public repo vs. 96 routes in the local-model dev checkout. The
+ in the public repo vs. 102 routes in the local-model dev checkout. The
desktop client's README describes surfaces (receipts, plugins, memory,
- studio, attest) that the public engine repo does not yet expose — an
+ studio, attest) that the public engine repo does not yet expose -- an
integrity risk for exactly the audience this portfolio courts.
- Distribution is GitHub-releases-only; no product page exists outside the
portfolio site; Windows-only installer today (macOS/Linux runners exist in
@@ -92,7 +92,7 @@ fold.
## Adjacent properties on the shelf
Private repos `harpercompliance-site` and `sendmyletter-site` (both touched
-2026-07-15) indicate property experiments beyond the portfolio — the compliance
+2026-07-15) indicate property experiments beyond the portfolio -- the compliance
name in particular aligns with vertical B and the owner's GRC documentation
history. Not assessed (private, out of session scope); noted as existing
assets when a domain strategy is chosen.
diff --git a/docs/deck/flywheel-telos-deck.html b/docs/deck/flywheel-telos-deck.html
index 9a8408c..df7f202 100644
--- a/docs/deck/flywheel-telos-deck.html
+++ b/docs/deck/flywheel-telos-deck.html
@@ -1,4 +1,4 @@
-Flywheel · Project Telos — The Whole Workflow
+Flywheel · Project Telos -- The Whole Workflow