Hermes Profiles

Curated Hermes Agent profiles for specialist swarms. Each profile packages a role identity (SOUL.md), skill dependencies, and configuration for a specific architecture/engineering capability that can be deployed as a Hermes agent.

These profiles are opinionated — they use Hermes-native patterns:

• Artifact-pyramid output format (progressive disclosure, path-as-handoff)
• skill-based methodology loading (skill_view → load references)
• Hermes kanban for multi-agent orchestration
• Hermes profile system for role isolation

The profiles (SOUL.md + profile.yaml) are Hermes-specific and unlikely to work in other harnesses. But the skills — the methodology directories in skills/ with their SKILL.md + references/ structure — follow the Agent Skills open standard adopted by Claude Code, Codex CLI, Cursor, Gemini CLI, OpenClaw, GitHub Copilot, Windsurf, and 20+ other coding agents. Those should work anywhere.

Repository Structure

hermes-profiles/
├── skills/
│   ├── architecture/
│   │   ├── adr-authoring/
│   │   ├── arc42-context/
│   │   ├── architect-pyramid/
│   │   └── c4-diagramming/
│   ├── artifact-pyramids/
│   ├── backend-engineering/
│   ├── brand-designer/
│   ├── chief-of-staff-methodology/
│   ├── contribution-pipeline/
│   ├── copy-editor-methodology/
│   ├── curation-methodology/
│   ├── data-architect/
│   ├── data-engineering/
│   ├── data-scientist/
│   ├── debugging-methodology/
│   ├── docker-management/
│   ├── editor-methodology/
│   ├── editor-review-methodology/
│   ├── editorial-methodology/
│   ├── executive-methodology/
│   ├── financial-modeling/
│   ├── frontend-engineering/
│   ├── go-to-market/
│   ├── implementation-planning/
│   ├── kanban-guru/
│   ├── legal-strategy/
│   ├── mermaid-diagrams/
│   ├── ml-engineering/
│   ├── opensource-contributions/
│   ├── operational-design/
│   ├── orchestration-methodology/
│   ├── org-design/
│   ├── platform-engineering/
│   ├── product-methodology/
│   ├── product-strategy/
│   ├── qa-methodology/
│   ├── research-methodology/
│   ├── researcher-workflow/
│   ├── review-methodology/
│   ├── sdd-authoring/
│   ├── sdd-review/
│   ├── sdd-verification/
│   ├── sdd-work-decomposition/
│   ├── seo-audit/
│   ├── seo-content-optimization/
│   ├── security-audit-methodology/
│   ├── site-reliability-engineering/
│   ├── software-architecture-analysis/
│   ├── strategy-frameworks/
│   ├── systematic-debugging/
│   ├── tailscale/
│   ├── technical-documentation/
│   ├── technology-radar/
│   ├── traefik/
│   ├── ux-methodology/
│   ├── verification-methodology/
│   └── wonderer-methodology/
└── profiles/
    ├── backend-engineer/             ← API implementation, service logic, database access
    ├── brand-designer/               ← Brand identity, visual systems
    ├── copy-editor/                  ← Line-level editing, proofreading
    ├── curator/                      ← Knowledge management, atomic notes
    ├── data-architect/               ← Data modeling, pipelines, governance
    ├── data-engineer/                ← Database ops, ETL, migrations, data quality
    ├── data-scientist/               ← Stats, causal inference, ML
    ├── debugger/                     ← Root cause analysis, error diagnosis
    ├── editor/                       ← Structural editing, argument coherence
    ├── frontend-engineer/            ← UI components, state management, API integration, performance
    ├── implementation-planner/       ← Work breakdown, critical path
    ├── kanban-strategist/            ← Flow optimization, WIP calibration
    ├── ml-engineer/                  ← Model training, fine-tuning, evaluation, deployment
    ├── orchestrator/                 ← Task decomposition, specialist routing
    ├── oss-contributor/              ← Open source contribution workflows
    ├── platform-engineer/            ← CI/CD, IaC, container orchestration, service networking
    ├── product-manager/              ← Specs, prioritization, strategy
    ├── qa-engineer/                  ← Test strategy, automation, quality gates
    ├── researcher/                   ← Deep investigation, evidence synthesis
    ├── reviewer/                     ← Code/architecture review, quality gates
    ├── security-engineer/            ← Threat modeling, vulnerability assessment, security architecture
    ├── seo-specialist/               ← Search optimization, metadata
    ├── site-reliability-engineer/    ← SRE: reliability engineering, incident command, observability
    ├── spec-driven-development/      ← SDD: spec authoring, task decomposition, verification, phase-gate review
    ├── technical-architect/          ← Systems architecture: C4 + ADRs + arc42
    ├── technical-writer/             ← API docs, READMEs, AGENTS.md, developer guides
    ├── ux-designer/                  ← User journeys, accessibility
    ├── verifier/                     ← Pass/fail gatekeeping, evidence
    ├── wonderer/                     ← Lateral exploration, overlooked angles, adjacent possibilities
    └── writer/                       ← Drafting, voice, narrative flow
    │
    │   ── C-Suite Profiles ─────────────────────────────────────────
    ├── ceo/                          ← Vision-to-strategy translation, capital allocation
    ├── cto/                          ← Technology strategy, architecture governance
    ├── cfo/                          ← Financial modeling, budget enforcement
    ├── coo/                          ← Operational design, execution infrastructure
    ├── cpo/                          ← Product strategy, roadmap, market fit
    ├── cmo/                          ← Go-to-market, brand, customer acquisition
    ├── chief-of-staff/               ← Leader capacity amplification, gatekeeping, coordination
    ├── clo/                          ← Legal strategy, regulatory risk, IP
    └── chro/                         ← Org design, talent strategy, culture

Profiles symlink to the shared skills/ directory, so one copy of each skill serves every profile. Git tracks symlinks by reference, not by duplicating content.

Using a Profile

# Clone the repo
git clone https://github.com/magnus919/hermes-profiles.git ~/hermes-profiles

# Symlink the profile you want into ~/.hermes/profiles/
# (39 profiles available — pick one)
ln -s ~/hermes-profiles/profiles/researcher ~/.hermes/profiles/

# Switch to profile (skills are bundled — no separate install needed)
hermes --profile researcher

Each profile's profile.yaml lists its required skills and any Hermes-specific configuration.

Contributing

Open an issue or PR. Profiles should:

• Have a clear, single responsibility
• Include a SOUL.md with first principles and output contract
• List skill dependencies explicitly
• Be Hermes-native (artifact-pyramid output, skill-based methodology loading)
• Not depend on agent-specific infrastructure (council, cashew, etc.) — these must work on a vanilla Hermes install

License

MIT