richfrem/os-architect
plugins/agent-agentic-os/skills/os-architect/SKILL.md
SME-facing front-door skill for Agentic OS ecosystem evolution. Invokes the os-architect interview flow: classifies intent, audits existing capabilities, proposes evolution path (orchestrate / update / create), and dispatches work. Use when evolving plugins, skills, or agents — whether applying a new pattern, setting up an improvement lab, filling a capability gap, or coordinating multiple loops.
$ install --global
skillsauthnpx skillsauth add richfrem/agent-plugins-skills os-architectInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Apr 27, 2026, 7:18 AM42.9s2 files scanned
SKILL.md
- name:
- os-architect
- plugin:
- agent-agentic-os
- description:
- >
- interview flow:
- classifies intent, audits existing capabilities, proposes evolution path
- model:
- inherit
- color:
- purple
- tools:
- ["Bash", "Read", "Write"]
Role
os-architect is the single entry point to the Agentic OS evolution ecosystem. The user
invokes it when they want to evolve or build anything in the agent/skill/plugin ecosystem.
It interviews, audits, and routes — never implements directly. The full behavior spec lives
in agents/os-architect-agent.md.
How to Invoke
/os-architect
No arguments needed. Start with a plain-language description of what you want to evolve.
What It Does
- Phase 1 — Intent Interview: classifies the request into one of 5 evolution categories
- Phase 2 — Ecosystem Audit: verifies what capabilities exist vs what's missing
- Phase 3 — Proposal + Dispatch: proposes Path A/B/C and dispatches via the user's CLI tools
Dispatch Paths
| Path | When | Mechanism |
|------|------|-----------|
| A+ — No Action | audit shows full match + all patterns present | tell user, no dispatch |
| A — Orchestrate | capability exists, current | route to existing agent/skill + run_agent.py |
| B — Update | capability exists, outdated/incomplete | os-evolution-planner writes plan + prompt → dispatch + optional improvement loop |
| C — Create | gap confirmed | create-sub-agent scaffold → os-evolution-planner plan + prompt → eval lab → evals HARD-GATE → os-architect-tester validates |
Related Capabilities
| Skill / Agent | Purpose |
|---------------|---------|
| os-evolution-planner skill | Called for Path B/C: writes structured task plan + Copilot CLI delegation prompt |
| os-architect-tester agent | Validates os-architect correctness via scenario simulation — run after any changes |
| improvement-intake-agent | Called for Category 3 (Lab Setup): configures the skill improvement run |
| create-sub-agent | Called for Path C (Gap Fill): scaffolds the new agent/skill file |
Gotchas
- Invoked without clear intent: If the user says only "help me" or "I don't know where to start", run Phase 1 open question first — do not assume intent category.
- PATH detection gives false positive:
which ghreturns a result but Copilot CLI is not configured. Always confirm tool availability with the user before using a detected tool for dispatch. - Path C evals HARD-GATE is non-negotiable: The evals review gate before the first improvement loop must not be skipped even if the user says "just run it." Bad evals waste improvement loop compute.
Smoke Test
- Given "I found a new browser harness pattern and want to apply it to my skills" → classifies as Category 1 (Pattern Abstraction), proposes Path A or B, within 2 turns
- Given "I need an agent that audits plugin evals for staleness, it doesn't exist yet" → classifies as Category 4 (Gap Fill), proposes Path C, surfaces eval HARD-GATE
- Given "run 50 iterations on my os-eval-runner skill" → classifies as Category 3 (Lab Setup), routes to improvement-intake-agent, asks about CLI tools before proposing dispatch strategy
- Given "I want to explore improving how we generate agents and maybe create one" → agent identifies Low confidence (Cat 3 + Cat 4 overlap), asks clarifying question before proceeding to Phase 2
Related Skills
richfrem/task-agent
data-ai
VerifiedTrustedCommunity
Task management agent. Auto-invoked for task creation, status tracking, and kanban board operations using Markdown files across lane directories. V2 enforces Kanban Sovereignty constraints preventing manual task file edits.
3SKILL.mdUpdated Jun 9, 2026
richfrem/symlink-manager
development
VerifiedTrustedCommunity
Create, audit, repair, and document cross-platform symlinks that work correctly on both Windows and macOS/Linux. Use this skill whenever the user mentions symlinks, symbolic links, junction points, .gitconfig symlinks, broken links after git pull, cross-platform path issues, or needs help with ln -s equivalents on Windows. Also trigger when the user reports that files are missing or wrong after switching between Mac and Windows machines using Git. This skill solves the common problem where symlinks committed on macOS show up as plain text files on Windows (and vice versa) because of Git's core.symlinks setting or missing Developer Mode / elevated permissions. **IMPORTANT FOR WINDOWS USERS:** Developer Mode must be enabled before creating symlinks. Without it, Git will check out symlinks as plain-text files or hardlinks, breaking cross-platform workflows.
3SKILL.mdUpdated Jun 9, 2026
richfrem/red-team-bundler
development
VerifiedTrustedCommunity
Interactively prepares a targeted Red Team Review package. It conducts a brief discovery interview to determine the threat model, generates a strict security auditor prompt, compiles a manifest of relevant project files, and bundles them into a single Markdown artifact or ZIP archive ready for an external LLM (like Grok, ChatGPT, or Gemini) or a human reviewer.
3SKILL.mdUpdated Jun 9, 2026
richfrem/optimize-context
tools
VerifiedTrustedCommunity
Reduces AI agent context bloat across three dimensions: (1) duplicate skill deduplication — clears stale agent directory copies since the IDE already reads from plugins/ directly; (2) instruction file optimization — rewrites CLAUDE.md, GEMINI.md, or .github/copilot-instructions.md to under ~80 lines, keeping only rules that directly change agent behaviour; (3) session token efficiency — guidance on cheap subagent delegation, context compounding across turns, and session hygiene. Trigger with "optimize context", "reduce context bloat", "deduplicate skills", "trim CLAUDE.md", "trim GEMINI.md", "fix my context usage", "why are my skills loading twice", "how do I reduce token usage", or "clean up agent directories".
3SKILL.mdUpdated Jun 9, 2026