Skip to content
Back to skills

Aid Update Architecture

ASecurity

Revise the project's build-and-shape (C1) Knowledge Base document -- components and their responsibilities, boundaries, interactions, and the invariants a change must not break -- plus any previously created outputs you name. Use this skill when the architecture document already exists and part of it needs revising. Requires no design seed: the change you state in the run is a sufficient input. Reads and consumes an architecture seed when one is present in `.aid/design/`. When the C1 document...

  • 6 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added November 7, 2026
toolsgoshellbashdocumentation

Works with

  • terminal

Security analysis

A100/100

Scanned November 7, 2026

npx -y skills add AndreVianna/aid-methodology --skill aid-update-architecture --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Aid Update Architecture?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Aid Update Architecture
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/andrevianna-aid-update-architecture/badge)](https://www.skillsdirectory.com/skills/andrevianna-aid-update-architecture)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: aid-update-architecture
description: >
  Revise the project's build-and-shape (C1) Knowledge Base document -- components and their
  responsibilities, boundaries, interactions, and the invariants a change must not break --
  plus any previously created outputs you name. Use this skill when the architecture
  document already exists and part of it needs revising. Requires no design seed: the change
  you state in the run is a sufficient input. Reads and consumes an architecture seed when
  one is present in `.aid/design/`. When the C1 document does not yet exist, routes to
  `/aid-create-architecture`. To write documentation ABOUT an architecture rather than
  revise the KB record, use `/aid-document-architecture`. Produced by the aid-architect
  agent and independently verified by aid-reviewer (full verify). Allocates a work-NNN
  folder.
allowed-tools: Read Glob Grep Bash Shell Terminal Write Edit Agent
argument-hint: "<change> -- what to revise in the architecture record"
---

# Update Architecture (revise the C1 document)

Bind **VERB=`update`**, **ARTIFACT=`architecture`** -- the `update` *stage* in the contract's
vocabulary -- then follow the shared contract at
`.aid/runtime/templates/design-lifecycle.md`. This skill binds that contract and restates
none of its rules.

- **Boundary vs `/aid-create-architecture`:** that skill realizes a seed and owns the
  creation path; this one is the maintenance verb. Where the destination is absent this skill
  routes there rather than standing one up itself.
- **Boundary vs `/aid-document-architecture`:** that skill writes documentation *about* an
  architecture; this one revises the Knowledge Base record.
- **Not a numbered pipeline phase**; does not route to `/aid-execute`.

State machine: **INTAKE -> UPDATE -> VERIFY -> PRESENT -> DONE**, as
`design-lifecycle.md § Skill shape` defines it. Print the `[State: NAME] -- {purpose}` entry
line on each state.

---

## State: INTAKE

1. **Require a stated change.** Empty argument -> ask one bootstrapping question ("What
   should change in the architecture record?") and wait.
2. **Allocate** through the Work Initiation Gate exactly as the contract's Allocation rule
   requires (`.agents/skills/aid-design/SKILL.md` INTAKE step 4 is the shape), with
   `initiator: aid-update-architecture`. `phase` is not driven.
3. **No seed is required.** The change stated in this run is a sufficient input, and this
   skill completes without a seed. **Read and consume one when present** at
   `.aid/design/architecture.md`, carrying its `## Current direction` into the destination and
   deleting it once realized.
4. **Resolve the destination by concern.** This skill binds concern **C1** (build & shape).
   Where a seed supplied a `## Destination`, use it; otherwise apply the concern rule here --
   resolve C1 against `.aid/settings.yml` `knowledge.doc_set`, falling back to the C1 row of
   `.aid/runtime/templates/kb-authoring/domain-doc-matrix.md` -- and **confirm the resolution
   with the user before writing**. Never resolve silently. C1's default set holds both
   `project-structure.md` and `architecture.md`; this skill targets the document describing
   the system's shape, never `project-structure.md`.
5. **Read the whole destination**, and note its `source:` field. Absent destination -> route
   to `/aid-create-architecture` and write nothing.
6. **Classify complexity** for the `aid-architect` dispatch (verifier tier >= producer).

**Advance:** UPDATE.

---

## State: UPDATE

1. **Apply only the change the user named in this run**, to the owned C1 content, per the
   content rules below.
2. **Ask, every run, which derived outputs to update alongside this one.** This step is
   unconditional -- it runs on every invocation, and the answer is **stored nowhere**: no
   frontmatter backlink, no manifest, no registry, and no state carried between runs. The
   question is asked afresh each time.
3. **Write no tracking metadata** into any output this run touches -- no `derived-from`, no
   `source-doc`, no `generated-by`, no `aid-tracked` field, and no skill-attribution line.
4. **`source: generated` refuses.** A registered build script owns that content.

**Advance:** VERIFY.

---

## State: VERIFY

**Full verify**, as `design-lifecycle.md` defines it. Not clean -> loop to UPDATE; the
3-cycle circuit breaker there escalates to IMPEDIMENT + `lifecycle: Blocked`.

**Advance:** PRESENT.

---

## State: PRESENT (hard stop -- the user decides)

Set `lifecycle: Paused-Awaiting-Input`. Present the revision, name every output touched, and
assert that a consumed seed is gone.

**Advance:** DONE.

---

## State: DONE

Set `lifecycle: Completed`, `updated` now, append a `## Lifecycle History` row.

---

## What this skill writes -- and must not

**Writes (C1):** components and their responsibilities; boundaries and what crosses them;
interactions and data flow; the invariants a change must not break; and what is deliberately
*not* a component.

**Must not write** -- the same boundaries `/aid-create-architecture` carries:

- **Framework or runtime versions** -- concern C0, the stack skills'.
- **Pipeline stages or environments** -- concern C8, the cicd skills'.
- **Rejected alternatives** -- concern D. They belong in the project's D document
  (`decisions.md` by default), never in the C1 doc.

## Frontmatter invariants

- The destination's `source:` is left **unchanged**, whatever it was -- neither verb rewrites
  it.
- `source: generated` **refuses**.
- `approved_at_commit:` is **never** written and never restamped -- it is generator-written by
  `/aid-discover` and `/aid-update-kb` on approval, and this skill is neither.
- `sources:` gains only what this run actually used.

## Write discipline

Read the destination whole, edit in place, and write back with everything outside the edited
range **byte-identical**. Never regenerate and never restructure the document. Adding a
`## ` section obliges updating that document's `## Contents` list in the **same** write.

## Conformance Lane

This skill changes **only what the user named in that run**. It does **not** resolve a flagged
Conformance-Lane divergence on its own -- reconciliation there is human by the lane's design --
so a flag the user did not point at survives this run untouched.

## Constraints

- **`phase` is not driven** by this skill.
- **Full verify**, per `design-lifecycle.md`.
- **No seed required; a present seed is consumed.**
- **Never stands up its destination** -- that path belongs to `/aid-create-architecture`.
- **Clean context**; verification always an `aid-reviewer` sub-agent dispatch.
- **Tracking:** write STATE `lifecycle` at every transition.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.