AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Decision

skill-alpacalabsllc-skills-for-architects-decision · by AlpacaLabsLLC

Capture a project decision as an ADR-style record in decisions/NNNN-slug.md — context, options considered, the call, consequences, status. Use when the user makes or reports a design, code, zoning, or procurement choice ("we're going with...", "the client decided..."), when an analysis surfaces options that force a choice, or to supersede an earlier decision. Records reasoning — current facts liv…

No reviews yet
0 installs
32 views
0.0% view→install

Install

$ agentstack add skill-alpacalabsllc-skills-for-architects-decision

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-alpacalabsllc-skills-for-architects-decision)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
2mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Decision? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

/decision — Project Decision Records

You capture decisions the way software teams capture ADRs (Architecture Decision Records) — except these are for actual architecture. One file per decision, numbered, in decisions/ at the project root. The record survives the email thread, the meeting, and the personnel change that would otherwise lose it.

Usage

/decision                          → interview for a decision in the air, then record it
/decision we're going with X       → record it, asking only for what's missing
/decision supersede 0003           → mark 0003 superseded, record its replacement
/decision list                     → show the index with statuses

Hard rules

  1. One decision per record. "Schemes B and code edition 2022" is two records.
  2. Never renumber, never delete. Numbers are permanent references. Wrong decisions get superseded, not removed — the reasoning trail is the point.
  3. Capture options honestly. A record with one option isn't a decision, it's an announcement. Ask what else was considered, even briefly.
  4. Keep the dossier index in sync. Every create/supersede updates the Decisions table in PROJECT.md (create the dossier via /project-dossier init first if it doesn't exist).
  5. Brevity. Context in 2–4 sentences, options as one line each, consequences as bullets. A record nobody reads records nothing.

Recording flow

  1. Find the next number: Glob decisions/*.md, take max + 1, zero-padded to 4 digits.
  2. Gather the pieces — from the conversation first, ask only for gaps (one grouped question): what was decided, what options were on the table, why, who made the call.
  3. Write decisions/NNNN-{kebab-slug}.md from the template.
  4. Update the Decisions index in PROJECT.md.
  5. Confirm with the path and a one-line restatement of the decision.

Template

# NNNN — {Decision title, stated as the choice made}

- **Status:** decided
- **Date:** {YYYY-MM-DD}
- **Deciders:** {who made the call}

## Context

{2–4 sentences: the situation that forced a choice. Link the analysis that surfaced it if one exists.}

## Options considered

1. **{Option A}** — {one line: what it would mean}
2. **{Option B}** — {one line}

## Decision

{The choice, and the load-bearing reason in 1–3 sentences.}

## Consequences

- {What this enables, costs, or constrains downstream}
- {What must now happen as a result}

Statuses: proposed (on the table, not yet made — record it so the options aren't lost), decided, superseded by NNNN.

On supersede

  1. Read the old record. Set its status to superseded by NNNN (naming the new record) — touch nothing else in it.
  2. Record the new decision normally; its Context section names what it replaces and why the reversal.
  3. Update both rows in the dossier index.

When analysis skills propose a decision

Skills in this marketplace surface choice points (zoning paths, code editions, GWP thresholds) and suggest running /decision. When that happens, the analysis is your Context — quote its numbers with their source rather than re-deriving them.

Collaboration

Records are plain files in the project folder — share them the way the project is already shared (git, Drive, Dropbox). Review happens in version control or in conversation; this skill imposes no workflow beyond the files.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.