← All modules
MOD-10
Autonomous Agent Foundry
Design agent crews with explicit roles, tool allowlists, memory and guardrails. Every run is traced with token, latency and cost accounting so autonomy stays auditable.

TRACE
100%
Starting at
$55
Tiers
3
Free Tier Sandbox
Free tier sandbox — Autonomous agent foundry
Visitor · 0 XP
Compose a multi-agent crew, wire up tools, and watch the plan-act-critique loop execute with live cost and latency accounting. Free tier runs simulated traces and exports the manifest — running real agents against your own data needs a paid tier.
next: Tinkerer
Model
FOUNDRY-70B
Agents
4
Tools
2
Est. cost
$0.0000
Tools granted
Execution trace
// run the crew to see the plan-act-critique loop
Tokens in
0
Tokens out
0
Wall time
0.00 s
Success est.
79%
| Tool | Calls | Avg ms | Cost |
|---|---|---|---|
| web.search | 0 | 420 | $0.0000 |
| db.query | 0 | 180 | $0.0000 |
Generated crew manifest
# Goke's Lab — agent crew manifest
version: 1
goal: "Find three suppliers for lithium anodes and draft outreach emails"
runtime:
model: FOUNDRY-70B
max_steps: 4
memory: vector + episodic
guardrails: [tool_allowlist, output_schema, budget_cap]
agents:
- role: planner
temperature: 0.4
- role: researcher
temperature: 0.4
- role: engineer
temperature: 0.4
- role: critic
temperature: 0.1
tools:
- web.search
- db.query
budget:
usd_cap: 0.05Tier Selection
Tier I
Recruit
$55
per month
- Simulated crew runs
- Tool allowlist + budget caps
- Crew manifest export
Tier II
Squad
$219
per month
- Live agents on your data
- Vector + episodic memory
- Webhook + scheduler triggers
- Full run tracing dashboard
Tier III
Command
$659
per month
- Unlimited crews and runs
- Private model routing
- Human-in-the-loop approvals
- Dedicated agent engineer