riskmandate.ai/admin·the site·the library·this page as markdownnoindex · not in the sitemap
Written by the agents that maintain this site and read by a person before it merges. Every count on this page is read off a file in the repository; where a page and a file disagree, the file is right.

admin / work / briefs/T11-consequences-and-assets.md

T11 — Consequences and assets: what follows when a capability meets what is in the deployment

A task brief, sized for one agent. The source is .claude/briefs/T11-consequences-and-assets.md; this page is that file, rendered.

Done v1.24.0 — built for oc433z3m and the template. See docs/briefs/direction__mvp-vault-and-the-reading-app.md and the v1.24.0 notes.

From: docs/briefs/direction__consequences-assets-and-the-vault-as-a-website.md §3 and §5 items 1 and 5. Size: a day. Touches: site/vaults/claude-gmail-connector/data/assets.json and data/consequences.json (new, authored), scripts/site/build-abp-vault.mjs (derives CONSEQUENCES.md, the routes-out table, the open-consequence view on the delta), site/vaults/_app/index.html (the What follows view under Evidence), site/vaults/_template/data/ (empty files with the schema, so every vault carries the layer), data/scenarios.json (two scenarios).

The task

The grant is a union of capabilities; the reader needs the consequences, each explicit and in the deployment's own words, with the asset that makes it real. Author the eleven consequences and five assets drafted in the brief for oc433z3m, with evidence tier, source and date on every one, and the open questions in the file. Derive CONSEQUENCES.md and a routes-out table (one row per exfiltration consequence: door, barrier, outlives the session, told not to). Recompute open consequences from the delta: every required capability in the excess or unstated, every required asset present or assumed. Add the two mandate scenarios the memo asks for.

Constraints

  • No score. A consequence has a kind, a barrier, an undo class and an evidence tier. Never a

    severity, a rating or a colour that ranks one above another.

  • Every claim is about the shape, none about Google. Facts, dates, sources. Suspension and

    sending limits are documented from Google's pages or left open; nothing is provoked (rule 4).

  • Explicit, second person, the mailbox's own words. "You can read any password-reset link

    that arrives in this mailbox", not "credential exposure".

  • present: assumed is its own word in every rendering; never silently present.
  • The schema is this site's extension, declared as such in the file (type, _what_this_is,

    provenance), until the model site sanctions one (Lab 03 ask).

  • Every existing grant row gets at least one child consequence or a line saying none was found.

Done means

  • data/assets.json and data/consequences.json exist in the Gmail vault and validate against

    the schema the build enforces; _template/ carries empty ones.

  • CONSEQUENCES.md is derived, lists the routes-out table, and the count of open consequences

    moves when mandate.json moves (checked by --check, like the delta).

  • The app shows What follows with the consequences grouped by kind, each with its requirements

    as links to the grant rows and assets, its evidence and its questions.

  • Two scenarios added; six existing unchanged.
  • npm run check green; the vault rebuilt; a note in RESEARCH-NEEDED.md for every open

    question the file raises.

An operations surface for the people and agents building this site, public because every page here is. It is not a selling page and nothing on it is for sale. The site · the library · every release · the repository · the script that wrote this page.