agentleFS
adezdevadezdev/blueprint/skills/architecture/SKILL.md

architecture

Skill0 starsChanged 3 months agoMITView on GitHub

Turn a requirements.md into a traceable architecture.md — components with stable C-ids, each declaring which R-ids it covers, plus decisions and risks. Use when the user says "/blueprint:architecture", "design this", "how should we build this", "turn the requirements into a design", or has a requirements.md and needs the technical shape before tasks are cut.

136 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use intelligent-routing, bash-linux and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • SKB3002Skill

    intelligent-routing

    Automatic agent selection and intelligent task routing. Analyzes user requests and automatically selects the best specialist agent(s) without requiring explicit user mentions.

    Also in 6 projects that use this

  • xenitV1Skill

    bash-linux

    Bash/Linux terminal patterns. Critical commands, piping, error handling, scripting. Use when working on macOS or Linux systems.

    Also in 8 projects that use this

  • xenitV1Skill

    red-team-tactics

    Red team tactics principles based on MITRE ATT&CK. Attack phases, detection evasion, reporting.

    Also in 8 projects that use this

  • SKB3002Skill

    nextjs-react-expert

    React and Next.js performance optimization from Vercel Engineering. Use when building React components, optimizing performance, eliminating waterfalls, reducing bundle size, reviewing code for performance issues, or implementing server/client-side optimizations.

    Also in 3 projects that use this

  • +8 more from xenitV1

The file

We split this skill into the rules it gives an agent

7 rules. 0 are common, 1 is distinctive, and 6 are very specific to this project.

  • 0

    Common

    Most of the 118 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read ${CLAUDE_PLUGIN_ROOT}/FORMAT.md first (or FORMAT.md at the repo root if you are working inside the blueprint repo itself).

    line 8
  2. Very specific:

    Stop before you start if ## Open questions still has unanswered Q<n> items — unanswered meaning no status: answered on the line.

    line 17
  3. Very specific:

    Record decisions as D<n>: only where a future reader would otherwise ask "why on earth".

    line 47
  4. Very specific:

    Prefer, in order: what the codebase already has → the standard library → an already-installed dependency → new code.

    line 63
  5. Distinctive:

    Every component names its files.

    line 82
  6. Very specific:

    No component without a covers: list.

    line 89
  7. Very specific:

    Do not write tasks here. Ordering and sequencing belong to /blueprint:tasks.

    line 91

“Common” is measured: kept by at least 6 in 10 of the 118 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Blueprint: architecture

Read ${CLAUDE_PLUGIN_ROOT}/FORMAT.md first (or FORMAT.md at the repo root if you are working inside the blueprint repo itself). It defines the ID grammar and the exact document shape. Do not improvise a variant.

Input

.blueprint/<slug>/requirements.md. Resolve the slug from the argument, or from the single existing .blueprint/*/requirements.md, or ask.

Stop before you start if ## Open questions still has unanswered Q<n> items — unanswered meaning no status: answered on the line. Surface them and ask, then have /blueprint:requirements record the answers before you design. Designing around an unresolved question produces a component that gets deleted.

What you do

  1. Read the existing codebase properly. Where does this feature attach? What patterns, helpers, and types already exist that this should reuse? The most common failure is an architecture that reinvents something living three files over. Grep before you design.
  2. Pick the approach and justify it against one real alternative. Two sentences on the alternative and why not. If you cannot name an alternative, you have not thought about it yet.
  3. Cut components at responsibility boundaries. One ### C<n> per unit with a single responsibility, a named file, and a concrete interface — real signatures, not prose. If a component's responsibility needs the word "and", split it. That applies to the heading too: a name like "Model and parser" describes two things even when the responsibility line underneath describes one, and the heading is what every later citation reads.
  4. Declare coverage on every component heading: covers: R1, R2.AC3. Then verify the other direction: every requirement in requirements.md must appear in at least one covers: list. A requirement no component covers is either a missed component or a requirement that should be dropped — say which, do not silently leave it.
  5. Record decisions as D<n> only where a future reader would otherwise ask "why on earth". Not every choice is a decision.
  6. Record risks — what could make this design wrong, and the cheapest early signal that it is.
  7. Write the file to .blueprint/<slug>/architecture.md.

Scope discipline

Design for the requirements in front of you. No extension points for requirements nobody wrote, no interface with one implementation, no configuration for a value that has one caller. If you believe a requirement is about to arrive that changes the shape, say so in ## Risks — do not build for it.

Prefer, in order: what the codebase already has → the standard library → an already-installed dependency → new code. A new dependency needs a line in ## Decisions justifying it.

Updating an existing architecture.md

Append new C-ids, never renumber. If implementation proved a component wrong, mark it status: dropped — <reason>, add the replacement as a new C-id, and list which tasks in tasks.md now point at a dropped component.

Done

Report: component count, the coverage check result (every R covered? which are not?), any new dependency you are proposing, and the top risk. Then:

Next: /blueprint:tasks <slug>

Rules

  • Every component names its files. "The auth layer" is not a component.
  • Two components may name the same file. Components are units of responsibility, not units of filesystem — splitting a small module three ways to satisfy a one-to-one rule produces files that exist only to import each other. When components share a file, say why in ## Approach. What they may never share is a responsibility.
  • Interfaces are code, not description.
  • No component without a covers: list. If it covers nothing, it should not exist.
  • Do not write tasks here. Ordering and sequencing belong to /blueprint:tasks.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.