This page summarizes how the wiseinvest-agents system is put together, based on README.md, AGENTS.md, runtime/, shared/integrations/mission_control.py, and docs under docs/runbooks/.
agents/<dept>/<id>/ implementing BaseAgent.handle(...). Their intent, tier, skills, triggers, and integrations are defined in catalog.yaml and agents/*/agent.yaml.runtime/) — common harness providing:
BaseAgent and AgentInput/AgentOutput (runtime/base.py).runtime/__main__.py) for python3 -m runtime list|run|run-all.runtime/loader.py) that maps agent IDs to classes.runtime/scheduler.py) that reads catalog.yaml triggers and dispatches agents.runtime/spawn.py) for Copilot-style sub-agent briefs.runtime/mc_bootstrap.py)../bin/wi wraps the runtime and scheduler into a single operator-friendly command surface (start/stop/status/logs/run/doctor/brief).shared/integrations/mission_control.py and shared/integrations/mc_cron.py publish agent/dispatch metadata into a Mission Control instance at MC_URL.settings.run_dir (default /tmp/wiseinvest-runs/).data/stocks/<TICKER>/stock_data.json maintained by the stock data pipeline.docs/ contains detailed instructions: running agents, scheduler semantics, Mission Control, architecture notes, and market research.These pieces together support the 10K-MAU growth plan in GROWTH_PLAN.md and the catalog in catalog.yaml.
catalog.yaml and agents/<dept>/<id>/ (config, prompts, code).python3 -m runtime ... or ./bin/wi ...) discovers agents and runs them by ID.runtime/scheduler.py, used via ./bin/wi start or scripts/start_scheduler.sh) reads catalog.yaml, maps triggers: fields onto cadences, and:
python3 -m runtime run <agent_id> as a subprocess.AgentInput (payload, trigger, run_id, context),handle(...) and returns an AgentOutput,settings.run_dir.For CLI and scheduler specifics, see Runtime, scheduler, and CLI. For projects and tasks in Mission Control, see Deploy & Mission Control.