Architect agent
Production AI agent skills for OpenClaw — extracted from a real personal AI OS. RAG memory, safety guards, cost tracking, architect agent and more.
npx -y skills add Mickos79/aria-skills --skill architect-agentAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
3.0 KB, 753 tokens by cl100k_base, as published. Nobody here has run it
architect-agent
Version: 1.0.0
Author: ARIA / Mikhail Kostenko
Model: Gemini Flash (via OpenRouter)
Requires: PostgreSQL (rag schema), OpenRouter API key
What it does
Classifies any user input (idea, note, task, project, meeting, research) and:
- Saves it to PostgreSQL
rag.itemswith tags, status, lifecycle tracking - For projects/research: creates full folder infrastructure automatically
- Links related existing items
- Returns a human-readable summary
Lifecycle
note → idea → research → project
Objects promote automatically based on signals (research started, code work begun, multi-step complexity detected).
Project infrastructure (auto-created for project type)
/aria-data/projects/{slug}/
CONTEXT.md — what it is, goals, architecture
TASKS.md — active/done/backlog tasks
DECISIONS.md — why we decided X (never lose reasoning)
RESEARCH.md — findings
MEMORY.md — auto-filled by archivist
logs/ — (code projects only)
.gitignore — (code projects only)
README.md — (code projects only)
/aria-data/backups/{slug}/
(backup target for project files)
Database schema
rag.items (
id UUID, type TEXT, status TEXT, title TEXT, body TEXT,
tags TEXT[], parent_id UUID, project_id UUID,
due_at TIMESTAMPTZ, folder_path TEXT, meta JSONB,
created_at TIMESTAMPTZ, updated_at TIMESTAMPTZ
)
rag.item_links (
from_id UUID, to_id UUID, relation TEXT
)
API endpoint
POST /architect
Body: { "text": "user input", "user_id": "optional" }
Response: {
"ok": true,
"item": { ...db row },
"classification": { type, title, tags, urgency, project_type, needs_infra },
"related": [ ...existing related items ],
"infra": { slug, projDir, backupDir, files } | null,
"summary": "human-readable result"
}
Classification types
| Type | Trigger signals | Infrastructure |
|---|---|---|
| note | quick thought, no action | none |
| idea | potential, may grow | none (until promoted) |
| task | concrete action, deadline | none |
| meeting | scheduled interaction | none |
| research | investigation needed | folder if complex |
| project | multi-step, code or business | full folder structure |
Integration
Called by the main model when it detects structured/complex input. Not called directly by the user — triggered automatically.
const { processInput } = require('./architect-agent');
const result = await processInput(userText, userId);
// result.summary → show to user
Files
architect-agent.js— main agent logicreferences/schema.sql— PostgreSQL schemascripts/install.sh— setup script
Install
npx clawhub@latest install aria/architect-agent
Then add to your aria-core index.js:
if (url === '/architect') {
const { processInput } = require('./architect-agent');
const result = await processInput(body.text, body.user_id);
respond(res, 200, { ok: true, ...result });
return;
}