The assistant

wardrik init creates an agent named assistant. Send it a request. It replies with a plan and the tool calls to carry it out. Each call runs after you approve it.

wardrik ask assistant "Tell me when https://github.blog/changelog/feed/ has a new post"

I'll set up a watcher on that feed. Plan:

• A worker, changelog-watch, checks the feed every 30 minutes. A check costs nothing until a
  new post appears.
• When one lands, it keeps a line about it, and I'll pass it on when you ask.

Waiting on a person for 1 call:
  ? agent_create  call_8d2ebbfe28aa4ee1a269dee2

wardrik approval approve assistant call_8d2ebbfe28aa4ee1a269dee2
wardrik approval deny assistant call_8d2ebbfe28aa4ee1a269dee2 --reason "Once an hour is enough."

On a denial, it receives your reason and proposes again.

See what runs

The workers it creates are ordinary agents in your org:

wardrik ls                              # each agent, and what it's doing
wardrik trigger ls changelog-watch    # its triggers, and when each runs
wardrik agent history changelog-watch # what it found

Or ask it: "What are you doing for me?" Ask it to change or stop work the same way. Each change is a tool call you approve.

Its file

~/.config/wardrik/agents/assistant.md

---
description: "Turns what people ask for into work that runs."
tools: [self/*, agents/*, agent_create, files, web_search, run_lua]
context: [MEMORY.md]
policy: {budget: {tokens: 2000000, per: day}}
idle_timeout: infinity
---
You are Wardrik, the one place the people of this org come to get things done. …

• agents/* reads, messages and redefines the org's agents. agent_create creates a worker.
• Reads run without approval. Every change needs approval.
• files gives it a workspace. MEMORY.md is loaded into every turn.
• No model: it uses defaults from config.yaml.
• Its workers inherit its policy and budget.
• web_search works once config.yaml sets brave.api_key.

init also writes the role and binding that allow it to act on the org's agents:

~/.config/wardrik/config.yaml

roles:
  workers: {rules: [{verbs: [create, define, read, ask], kinds: [agent]}]}
bindings:
  - {org: acme, role: workers, subjects: [agent:assistant]}

See Orgs, roles and tokens.

Change it

• The body is its instructions, followed by a playbook of patterns that work. Edit it, then wardrik reload. init never overwrites the file.
• To message workers without approval, allow the tool: policy: {allow: [agent_ask], ...}.
• Add channels: [slack] to use it from Slack.
• After it reads content from outside, every change needs approval, even if allowed.