Counsel Pack

P4 checklist (counsel verification)

Use this as a quick “is the record defensible?” scan. Each item points to the primary place to review.

ID	Check	Pack page
P4-1	Every material claim has at least one primary-source anchor (canonical URL + #:~:text AND Ctrl+F string).	`(Phase D mapping step)`
P4-2	Risk landscape lanes match cited evidence (no uncited narrative).	`landscape.html`
P4-3	Foreseeable misuse vectors have concrete counsel questions and mitigation levers tied to evidence.	`foreseeable_misuse.html`
P4-4	Anthropic distinctives page does not overstate; each distinctive is supported by an evidence pointer.	`anthropic_distinctives.html`
P4-5	Acceptance table aligns with counsel work steps and referenced risk controls.	`acceptance_table.html`
P4-6	No broken internal navigation or misleading links; pack is reviewable offline.	`index.html`

Definitions & scope

Purpose. The Evidence & Source Integrity (ESI) module is the pack’s verification spine: it defines how claims are traced to evidence, and how reviewers can audit (and challenge) every material statement.

Claim ID: a stable identifier (e.g., S1-001) used across pages, the Evidence Ledger, and the Casebook.
Material claim: a statement that could plausibly affect counsel’s risk assessment, posture, or recommendations.
Canonical URL: the preferred, copy/paste-safe URL to cite for that claim (avoid tracking links and unstable mirrors).
Ctrl+F anchor: one or more short strings expected to reliably locate the relevant passage on the canonical page/PDF.
Text fragment (#:~:text): optional deep-link target for supported browsers; use in addition to a Ctrl+F anchor.
Altitude: how “deep” the capture is in our pipeline (e.g., board → page_raw → stratosphere). Higher altitude generally means more durable verification artifacts.
Comprehensive Sources Registry: the single source of truth for all external URLs referenced anywhere in the pack.

Non-goal. ESI does not provide legal advice; it provides defensible traceability and review mechanics.

Primer: how to verify a claim

Start with the Claim ID (from any page) and locate its row in the Evidence Ledger.
Open the canonical URL and confirm it matches what the pack cites (no redirects to unrelated pages).
Use the Ctrl+F anchor(s) from the ledger row to find the relevant passage quickly.
Where appropriate, also validate a text fragment (#:~:text) target to reduce reviewer time.
Record outcomes: verified / ambiguous / missing / drifted. If drifted, capture what changed and when, and consider a re-harvest.
Cross-check the Comprehensive Sources Registry to ensure the canonical URL is registered and discover any related aliases or PDFs.

Availability & drift notes

Some domains may block automated fetching or rate-limit non-browser clients. For example, CourtListener returned 403 Forbidden in our automated fetch environment; reviewers should rely on normal browser access and/or harvested captures when available.

When you encounter access friction (paywalls, bot-blocks, expired PDFs), treat it as evidence risk: record it, and prefer higher-altitude captures or alternate canonical sources where defensible.

Governance: registry closure & transition guardrails

We treat “introducing a new external source” as a governed event. Pack builds enforce a closure step that regenerates the Comprehensive Sources Registry and audits that every external URL in any page appears there.

During masterplan transitions (and other “drift-prone” moments), a ritual-vault check surfaces relevant watchers (e.g., banner/nav unification, link-closure audits, and source-registry closure) so governance does not silently drop between phases.

Demo storyline (6–10 minute walkthrough)