Three lightweight checks run after every edit; research and maintenance agents are started for specific jobs; the owner controls direction and publication. The practice ladder's harness section explains the underlying design.
Standing agents
The builder
Writes · edits · deploysThe general-purpose coding agent (Devin) that drafts pages, runs research, edits prose, and deploys the site. The agent writing this sentence.
Role- Turns human direction into pages, edits, and deployments
- Verifies its own work: link resolution, citation integrity, served-page checks after every deploy
- Never merges bulk rewrites of the owner's pre-existing documents without asking
- Three lightweight audit hooks fire on every edit: prose quality, site structure, and newly added uncited facts
- Human approval for direction, architecture decisions, and publishing
Prose Auditor
Prose quality · hookDetects the statistical fingerprints of unedited AI prose and forces a context-aware rewrite before text ships.
What it audits- All Markdown and HTML prose in the package; parser strips markup and entities and skips code, CSS/JS, and source quotations
- Two severity classes: fix-on-sight (em-dashes, spaced en-dashes) and flag-for-judgment (~20 grammar and phrase tells, e.g.
moreover,delve,it's worth noting,not just X but Y, intensifiers) - Every occurrence gets a line and column; missing files are errors rather than silently ignored; flags may be deliberately kept after review
- Detection is deterministic and regression-tested; rewrites are done by the agent, in context, never by regex
Structural Auditor
Organization · navigation · hookKeeps the site's structure and navigational flow coherent as pages multiply.
What it audits- Nav completeness: every page lists every page, same order; the current page marked exactly once
- Header hierarchy: compact global masthead first, page-specific title in a separate hero, and former home-page chapters distributed across Strategy, Practice, and Sources with no redundant chapter bar (regression-tested)
- Link integrity: every internal link and #anchor resolves; no orphan pages
- Page structure: title, meta description, exactly one h1, footer back-link, heading hierarchy
- Judgment pass (per skill): purpose separation, next-step affordances, label honesty, placement of new content
Reference steward
Link rot · quarterlyGuards the citation apparatus against link rot. References are written to survive dead URLs (author, title, publisher, year on each); this agent makes sure they rarely have to.
What it audits- Every external link, with one controlled retry: only 404/410 count as dead; network/TLS/5xx remain unreachable and are rechecked
- Every run writes a timestamped comparison report and identifies status changes from the previous audit
- Wayback snapshots are age-checked; the resumable manifest is saved after every URL
- Local PDF copies use collision-resistant names, PDF-signature validation, SHA-256 checksums, byte counts, and atomic writes
Evidence Auditor
Fact integrity · tripwire + quarterlyVerifies that every factual claim on the site matches its cited source, and that no number stands without evidence. The audit that protects the site's credibility.
What it audits- Quantitative claims plus high-risk qualitative claims (first, only, largest, majority), with stable claim IDs and source line numbers
- Numbered citations resolve to their reference-list URLs; direct external links count as evidence pointers; quotations, nav, code, and reference entries are excluded
- A lightweight edit hook flags newly written uncited facts; the full audit fetches sources and compares figure, date, and framing
- Uncited claims are sourced, labeled as estimates/examples, softened, or removed; vendor-reported numbers keep their "as reported" framing
- Verdict classes: supported / partially supported / unsupported (fixed immediately) / unreachable (cross-checked against secondary coverage)
Ephemeral agents: the research fleet
The AI roadmap was researched by five parallel research agents, each spun up for one question and shut down when it reported. Their outputs were cross-checked, and every figure that made the page carries a numbered citation in the roadmap's reference list.
| Research agent | Its question | What it fed |
|---|---|---|
| Maturity frameworks | What staged AI-adoption models exist, and where is the industry actually at? | The industry-state table; the four-phase shape |
| Use cases and ROI | Which specialty-insurance AI use cases have measured results? | Phase 2–3 use cases; the economics section |
| Competitive landscape | What have named competitors and vendors actually shipped? | The competitor table; build-vs-buy guidance |
| Regulatory | What do regulators expect, by when, in which states? | The governance gates and the regulatory sequencing section |
| Implementation | Why do most AI programs stall, and what do the 7% do differently? | The failure-mode table; the 1:1 adoption rule; the exit gates |
Editing and publishing
The owner can edit every page through the private browser editor. Sveltia CMS reads the existing HTML through a custom adapter, commits the revised page to GitHub, and triggers the auditor workflow. Production deployment runs only after the suite passes.
The public editor contains no repository secret. GitHub authentication is supplied by the owner at sign-in, and Cloudflare Access can restrict the /admin path to an email allow-list.
How the harness maps to this site
| Harness layer | How this site uses it |
|---|---|
| Context engineering | House style, tone, and rewrite rules live in the skill files each agent reads before acting; agents see the rules, not vibes |
| Skills | The Prose Auditor and Structural Auditor are packaged procedures: versioned, testable, invocable on demand (/prose-auditor, /structural-auditor) |
| Hooks | Prose, structural, and lightweight evidence tripwires fire after every write/edit; full link archiving and source verification remain scheduled because they require network calls |
| Verification loops | Citation-integrity checks, link resolution, tag balance, and served-page 200 checks after every deploy |
| Human gate | Aaron sets direction, approves architecture and bulk changes, and decides what ships. Consolidations (like the August 2026 harness-into-practice-ladder merge) are owner decisions; the Structural Auditor enforces the rules after them. Slop is cheap; trust is engineered |