ossd-add-collection - SKILL.md Agent Skill

name: ossd-add-collection description: Use when creating a new collection of related projects in oss-directory — by ecosystem, grant round, thematic grouping, or external list

Add Collection

You help maintainers create new collections in the oss-directory repository. Collections are groups of related projects organized by ecosystem, grant program, or other meaningful categorization.

When to Use This Skill

Use this skill when:

Creating a new collection of related projects
Organizing projects by ecosystem (e.g., "Arbitrum Ecosystem")
Grouping grant recipients or program participants
Creating thematic collections (e.g., "DeFi Protocols")

Do not use this skill when:

Adding a single project (use add-project skill)
Updating an existing collection (use update-project skill)

Workflow Steps

0. Working Directory

All commands assume you're in the oss-directory root. If not already there: cd ~/GitHub/oss-directory

1. Gather Collection Information

Ask the user for (or extract from provided info):

Collection name (required) - Unique identifier, lowercase, hyphenated
- Examples: ethereum-crypto-ecosystems, arb-stip-1, optimism-retro-funding-3
Display name (required) - Human-readable name
- Examples: Ethereum (Crypto Ecosystems), Arbitrum STIP Round 1, Optimism Retro Funding 3
Description - Brief description of what this collection represents
Projects list (required) - Array of project names to include
- Must be existing project names from the directory

2. Determine File Location

Collections are stored in a flat directory:

Path: data/collections/[collection-name].yaml
Examples:
- ethereum-crypto-ecosystems.yaml
- arb-stip-1.yaml
- optimism-retro-funding-3.yaml

3. Verify Projects Exist

Before creating the collection, verify all project names exist:

# For each project name, verify it exists
find data/projects -name "[project-name].yaml"

If any projects don't exist:

Ask user if they want to add them first (use add-project skill)
Or remove them from the collection list

4. Create Collection YAML File

Required format:

version: 7
name: collection-name
display_name: Collection Display Name
projects:
  - project-one
  - project-two
  - project-three

With optional description:

version: 7
name: collection-name
display_name: Collection Display Name
description: Brief description of this collection
projects:
  - project-one
  - project-two
  - project-three

Important:

Always use version: 7
Collection name must match filename (without .yaml)
projects array must have at least 1 item
Project names must match exactly (case-sensitive)
Projects are typically listed alphabetically

5. Validate and Lint

Run validation in the oss-directory repo:

pnpm run validate
pnpm lint

If there are formatting errors, auto-fix with:

pnpm prettier:write

The validator will check:

All referenced projects exist
Schema compliance
No duplicate project entries

6. Commit Changes

Only commit after successful validation.

Commit message format:

Add [Collection Display Name] collection

or

Create [collection-name].yaml

Examples from recent commits:

Create dao-drops.yaml
Add Arbitrum STIP Round 1 collection

Git commands:

git add data/collections/[collection-name].yaml
git commit -m "Add [Collection Display Name] collection"

Do not push unless explicitly requested by the user.

Common Patterns

Ecosystem Collections

Collections of projects in a blockchain ecosystem:

version: 7
name: arbitrum-ecosystem
display_name: Arbitrum Ecosystem
description: Projects building on Arbitrum
projects:
  - aave
  - curve
  - gmx
  - uniswap

Grant/Funding Round Collections

Collections of grant recipients:

version: 7
name: optimism-retro-funding-3
display_name: Optimism Retro Funding Round 3
description: Recipients of Optimism Retro Funding Round 3
projects:
  - project-a
  - project-b
  - project-c

Thematic Collections

Collections grouped by category:

version: 7
name: defi-protocols
display_name: DeFi Protocols
description: Decentralized finance protocols
projects:
  - aave
  - compound-finance
  - curve
  - uniswap

Validation Checklist

Before committing, verify:

Filename matches collection name field exactly
version: 7 is set
Required fields present: version, name, display_name, projects
All project names in the list exist in the directory
No duplicate project entries
Projects are sorted alphabetically (recommended)
YAML is properly indented (2 spaces)
pnpm run validate passes
pnpm lint passes
No duplicate collection (check existing collections)

Common Issues

Referenced project doesn't exist:

Run: find data/projects -name "[project-name].yaml"
If not found, add the project first or remove from collection

Duplicate collection name:

Check existing collections: ls data/collections/
Choose a different name or append qualifier

Project name mismatch:

Project names are case-sensitive
Must match the name field in the project file exactly
Common error: using display name instead of project name

Formatting errors:

Run pnpm prettier:write to auto-fix
Ensure proper YAML indentation (2 spaces)

Finding Project Names

To find the correct project name (not display name):

# Search by keyword
grep -r "display_name: Keyword" data/projects/

# List all projects
find data/projects -name "*.yaml" -exec basename {} .yaml \;

# Read a specific project file
cat data/projects/[first-char]/[project-name].yaml

Finding Existing Projects from GitHub URLs

CRITICAL: When you have a list of GitHub repository URLs and need to find which oss-directory projects they map to, you MUST use the proper search strategy. Simply looking for filenames matching the org name is insufficient.

Why Simple Filename Search Fails

Many repos map to projects in non-obvious ways:

Different project names: defillama GitHub org → defi-llama.yaml project file
Specific repo mappings: argotorg/solidity → solidity-ethereum.yaml (not argotorg.yaml)
Umbrella organizations: Multiple repos like a16z-crypto/helios, a16z-crypto/magi → single a16z-crypto.yaml
Org name variations: transferwise GitHub org → wise.yaml project file

Proper Search Strategy (Priority Order)

When searching for a GitHub URL like github.com/ethereum/go-ethereum:

1. Search for the specific repo URL first:

grep -r "github.com/ethereum/go-ethereum" data/projects/

This finds projects that explicitly list this specific repository, even if the project name doesn't match the org.

2. If not found, search for the org URL:

grep -r "github.com/ethereum" data/projects/

This finds org-level projects that track all repos from that organization.

3. If still not found, try filename match:

find data/projects -name "ethereum.yaml"

This is the fallback for when the project file directly matches the org name.

Real-World Examples

Example 1: Specific Repo Mapping

# Looking for: github.com/argotorg/solidity
grep -r "github.com/argotorg/solidity" data/projects/
# Returns: data/projects/s/solidity-ethereum.yaml

# This would FAIL:
find data/projects -name "argotorg.yaml"
# Returns: (nothing)

Example 2: Org-Level Project

# Looking for: github.com/succinctlabs/sp1
grep -r "github.com/succinctlabs/sp1" data/projects/
# Returns: (nothing - specific repo not listed)

grep -r "github.com/succinctlabs" data/projects/
# Returns: data/projects/s/succinctlabs.yaml
# (org-level project tracking all succinctlabs repos)

Example 3: Name Variation

# Looking for: github.com/defillama/chainlist
find data/projects -name "defillama.yaml"
# Returns: (nothing)

grep -r "github.com/defillama" data/projects/
# Returns: data/projects/d/defi-llama.yaml
# (project name uses hyphenated version)

Systematic Batch Search

When processing multiple GitHub URLs, use this approach:

# Create a list of repos to search
repos=(
  "ethereum/go-ethereum"
  "argotorg/solidity"
  "defillama/chainlist"
)

# For each repo, search in priority order
for repo in "${repos[@]}"; do
  org=$(echo "$repo" | cut -d'/' -f1)
  echo "Searching for: $repo"

  # Try specific repo first
  result=$(grep -l "github.com/$repo" data/projects/*/*.yaml 2>/dev/null)

  if [ -n "$result" ]; then
    echo "  Found (specific): $result"
  else
    # Try org-level
    result=$(grep -l "github.com/$org\"" data/projects/*/*.yaml 2>/dev/null)
    if [ -n "$result" ]; then
      echo "  Found (org): $result"
    else
      # Try filename
      result=$(find data/projects -name "$org.yaml" 2>/dev/null)
      if [ -n "$result" ]; then
        echo "  Found (filename): $result"
      else
        echo "  NOT FOUND"
      fi
    fi
  fi
done

Important Notes

Always search content first, filenames last: The project file content is authoritative
Check both specific repos and org-level: Some projects track individual repos, others track entire orgs
Handle umbrella organizations: Projects like a16z-crypto intentionally aggregate multiple repos that could be standalone
Document your mappings: Create a mapping file showing which repos map to which projects for transparency
Case sensitivity matters: Project names and GitHub URLs are case-sensitive

When Creating Collections from External Lists

When you receive a list of GitHub repositories (e.g., grant recipients, ecosystem projects):

Search systematically using the priority order above
Track your results in a mapping file (helpful for debugging)
Identify missing projects that need to be created
Verify umbrella orgs - some repos should share a project file, not have individual ones
Ask before splitting - if repos are under a single org, confirm whether to create org-level or repo-specific projects

Example Collection File

version: 7
name: base-ecosystem
display_name: Base Ecosystem
description: Projects building on Base
projects:
  - aerodrome
  - base-org
  - coinbase
  - morpho-org
  - uniswap

Adding Projects to Existing Collections

If you need to add projects to an existing collection, use the update-project skill instead.

Resources

Within the oss-directory repository:

Collection schema: src/resources/schema/collection.json
Example collections: data/collections/
All projects: data/projects/