ProjectMnemosyne is a skills marketplace for the HomericIntelligence agentic ecosystem. Named after Mnemosyne, the Greek goddess of memory, this repository serves as the collective memory where team learnings are preserved and made searchable.
- Claude Code CLI installed
- Access to HomericIntelligence/ProjectMnemosyne (private repository)
Register the private repository directly.
From terminal:
claude plugin marketplace add https://github.com/HomericIntelligence/ProjectMnemosyneOr inside a Claude Code session:
/plugin marketplace add https://github.com/HomericIntelligence/ProjectMnemosyne
If you have the repository cloned locally:
git clone https://github.com/HomericIntelligence/ProjectMnemosyne.gitFrom terminal:
claude plugin marketplace add /path/to/ProjectMnemosyneOr inside a Claude Code session:
/plugin marketplace add /path/to/ProjectMnemosyne
From terminal:
claude plugin marketplace listOr inside a Claude Code session:
/plugin marketplace list
These commands work inside any Claude Code session after installation. All skill discovery and installation is handled automatically through the /advise and /learn commands.
/advise <your goal or question>
Claude will search the marketplace for relevant prior learnings and return:
- What worked in similar situations
- What failed and why (critical!)
- Recommended parameters and configurations
/learn
After an experiment or debugging session, capture your learnings as a new skill. Claude automatically:
- Analyzes your entire session conversation
- Extracts successes, failures, and parameters
- Creates a branch and opens a PR with a new skill
Auto-trigger: On /exit or /clear, you'll be prompted to save learnings.
skills/
├── <name>.md # Flat skill files with YAML frontmatter
├── <name>.notes.md # (Optional) Additional session context
└── ...
plugins/
└── tooling/
└── mnemosyne/ # /advise and /learn commands
Each skill is a flat markdown file with YAML frontmatter:
skills/<name>.md # Main skill file with YAML frontmatter + markdown content
skills/<name>.notes.md # (Optional) Additional context from development session
See marketplace.json for the complete searchable index of available skills. The /advise command will automatically search and retrieve relevant skills for your queries.
- Complete an experiment or debugging session
- Run
/learn - Follow the prompts to categorize and name the skill
- PR is created automatically
- Copy
templates/skill-template.mdtoskills/<name>.md - Fill in YAML frontmatter (name, description, category, date, version)
- Fill all required markdown sections
- Include "Failed Attempts" table (required!)
- Create PR
- Overview table: Date, objective, outcome
- When to Use: Specific trigger conditions
- Verified Workflow: Step-by-step that worked
- Failed Attempts: What didn't work and why (REQUIRED)
- Results & Parameters: Copy-paste ready configs
- References: Links to issues, docs
All PRs are validated by CI:
- YAML frontmatter has required fields (name, description, category, date, version)
- All required markdown sections are present
- Failed Attempts section is present with proper table format
- Description is specific (20+ chars)
- Category is valid
Run validation locally:
python3 scripts/validate_plugins.py| Project | Purpose |
|---|---|
| Odysseus | Ecosystem orchestrator and architecture documentation |
| ProjectMnemosyne | Knowledge, skills, and memory (this repo) |
| ProjectHephaestus | Shared utilities and foundational tools used across the ecosystem |
| Project | Purpose |
|---|---|
| Myrmidons | GitOps agent provisioning — agent definitions as code, reconciliation against ai-maestro API |
| AchaeanFleet | Container images for the heterogeneous agent mesh — base images, Dockerfiles, Compose, Nomad/Dagger CI |
| Project | Purpose |
|---|---|
| ProjectKeystone | DAG execution and task coordination |
| ProjectHermes | Webhook-to-NATS messaging bridge |
| ProjectTelemachy | Workflow engine |
| ProjectProteus | CI/CD pipeline management |
| ProjectArgus | Observability and monitoring |
| Project | Purpose |
|---|---|
| ProjectOdyssey | Training framework written in Mojo |
| ProjectScylla | Testing, optimization, and resilience evaluation |
Note: Skills produced by any of the above repositories can be contributed to ProjectMnemosyne via
/learnso that learnings are shared across the ecosystem.
Mnemosyne embodies memory, continuity, and the preservation of wisdom. This repository ensures that learnings compound over time - every experiment, debugging session, and architectural decision becomes searchable knowledge for the team.
The most valuable section in any skill is Failed Attempts - knowing what didn't work saves more time than knowing what did.
If you use ProjectMnemosyne in your research or work, please cite:
@misc{projectmnemosyne2026,
title={ProjectMnemosyne: A Skills Marketplace for HomericIntelligence},
author={{HomericIntelligence Team}},
year={2026},
note={Skills marketplace and collective memory system},
url={https://github.com/HomericIntelligence/ProjectMnemosyne}
}