Skip to main content

​
PROJECT.md — Fleet-Wide Context

Shared across all agents. Loaded into every agent’s system prompt. 
# PROJECT.md

## What We're Building
SaaS platform for automated lead qualification

## Current Priority
Ship the email personalization pipeline this week

## Hard Constraints
- Budget: $50/day total across all agents
- No cold outreach to .edu or .gov domains

​
config/mesh.yaml — Framework Settings

mesh:
  host: "0.0.0.0"
  port: 8420

llm:
  default_model: "openai/gpt-4o-mini"
  embedding_model: "text-embedding-3-small"
  max_tokens: 4096
  temperature: 0.7

​
config/agents.yaml — Agent Definitions

Created automatically by openlegion setup or openlegion agent add. 
agents:
  researcher:
    role: "research"
    model: "openai/gpt-4o-mini"
    skills_dir: "./skills/research"
    system_prompt: "You are a research specialist..."
    resources:
      memory_limit: "512m"
      cpu_limit: 0.5
    budget:
      daily_usd: 5.00
      monthly_usd: 100.00

​
Agent Config Fields

FieldTypeDescription
rolestringAgent’s role identifier
modelstringLLM model in LiteLLM format (e.g., openai/gpt-4o-mini, anthropic/claude-sonnet-4-20250514)
skills_dirstringPath to custom skills directory
system_promptstringAgent’s system prompt
resources.memory_limitstringContainer memory limit (default: 512m)
resources.cpu_limitfloatCPU quota (default: 0.5)
budget.daily_usdfloatDaily LLM spend cap
budget.monthly_usdfloatMonthly LLM spend cap
mcp_serverslistMCP tool servers

​
config/permissions.json — Agent Permissions

Per-agent access control with glob patterns for blackboard paths and explicit allowlists for messaging, pub/sub, and API access. 
{
  "researcher": {
    "can_message": ["orchestrator"],
    "can_publish": ["research_complete"],
    "can_subscribe": ["new_lead"],
    "blackboard_read": ["tasks/*", "context/*"],
    "blackboard_write": ["context/prospect_*"],
    "allowed_apis": ["llm", "brave_search"]
  }
}

​
Permission Fields

FieldTypeDescription
can_messagelistAgents this agent can send messages to
can_publishlistPub/sub topics this agent can publish to
can_subscribelistPub/sub topics this agent can subscribe to
blackboard_readlistGlob patterns for readable blackboard keys
blackboard_writelistGlob patterns for writable blackboard keys
allowed_apislistExternal APIs accessible through the vault
Default policy is deny — agents can only do what’s explicitly permitted.

​
config/workflows/*.yaml — Workflow Definitions

See Workflows for the full specification.

​
.env — API Keys

Managed automatically by openlegion setup and openlegion channels add. Can also be edited directly: 
OPENLEGION_CRED_ANTHROPIC_API_KEY=sk-ant-...
OPENLEGION_CRED_MOONSHOT_API_KEY=sk-...
OPENLEGION_CRED_OPENAI_API_KEY=sk-...
OPENLEGION_CRED_BRAVE_SEARCH_API_KEY=BSA...
OPENLEGION_CRED_TELEGRAM_BOT_TOKEN=123456:ABC...
OPENLEGION_CRED_DISCORD_BOT_TOKEN=MTIz...

# Log format: "json" (default) or "text" (human-readable)
OPENLEGION_LOG_FORMAT=text
All OPENLEGION_CRED_* variables are loaded by the credential vault and never exposed to agent containers.

​
Project Structure

config/
├── mesh.yaml                           # Framework settings
├── agents.yaml                         # Agent definitions (per-project)
├── permissions.json                    # Per-agent ACLs
├── cron.json                           # Persistent cron jobs
├── watchers.yaml                       # File watcher definitions
└── workflows/                          # Workflow YAML definitions

PROJECT.md                              # Fleet-wide context
.env                                    # API keys (gitignored)