Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/magnusfroste/flowwink/agent-setupgit clone --depth 1 https://github.com/magnusfroste/flowwinkWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00039 | $0.01192 |
| Opus 5 | $0.00019 | $0.00596 |
| Sonnet 5 | $0.00008 | $0.00238 |
| Haiku 4.5 | $0.00004 | $0.00119 |
Grade A, and why
agent-setup scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 124 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent Setup
How an external agent (OpenClaw, Hermes, ClawThree, Claude Desktop, custom MCP claw) goes from "nothing" to "running a department on this FlowWink site".
This is the flow we used to seed the clawable.org agentic handbook simulations — Claude Code acted as game master (spelledare), generated invites, assigned roles to OpenClaw peers, then observed and reset.
1. Generate an invite
/admin/federation → Agent Invites
Pick one of the 5 operator missions (all category: 'operator'):
| Mission | When to use |
|---|---|
| Full Operator | One peer owns everything (no department split) |
| Growth Operator | Lead gen, pipeline, conversion |
| Commerce Operator | Orders, inventory, fulfillment |
| HR Operator | Hiring, contracts, onboarding |
| Finance Operator | Invoicing, expenses, BAS 2024 accounting |
| Custom Mission | Write your own instructions |
Each invite generates a fwk_* MCP key (visible under /admin/developer → MCP Keys) and renders a copy-pasteable bootstrap payload — base URL,
bearer token, two protocol options (native MCP / REST facade), key resources,
and the mission text. See agent_invite.md for the
full template.
For department-specialist peers, also point them at the matching
department playbook and the ?groups=<dept> filter.
2. Hand off to the peer
Paste the invite payload into the peer's chat / config. The peer should:
GET /rest/resources/briefing— identity + modules + health in one callGET /rest/tools?groups=<dept>(ortools/list) — discover live catalog- Begin the operating cadence described in the mission
No further setup on the FlowWink side. The MCP key authorises everything;
inactive modules are auto-hidden from the catalog
(mem://architecture/mcp-module-aware-filtering).
3. Game Master pattern (spelledare)
For demos, beta tests, training scenarios, or proving a process end-to-end, use the game master loop. One human (or Claude Code) orchestrates; multiple peers play roles.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 124 lines · 39 tokens per session scan A df021d64f570
agent-setup is an agent published in the GitHub repository magnusfroste/flowwink (21 stars, last pushed 2d ago), licensed MIT. It adds 39 tokens to every session and 1,192 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
stage-20-lodging
Agent "stage-20-lodging" from TelivityAI/otaip, covering stage 20 -- lodging agents, agent 20.1 -- hotel search aggregator, agent 20.2 -- property deduplication, agent 20.3 -- hotel content normalization and agent 20.4 -- hotel rate comparison.
stage-0-reference
Agent "stage-0-reference" from TelivityAI/otaip, covering stage 0 -- reference data agents, agent 0.1 -- airport/city code resolver, agent 0.2 -- airline code & alliance mapper, agent 0.3 -- fare basis code decoder and agent 0.4 -- class of service mapper.
stage-9-platform
Agent "stage-9-platform" from TelivityAI/otaip, covering stage 9 -- platform agents, agent 9.1 -- orchestrator, agent 9.2 -- knowledge retrieval, agent 9.3 -- monitoring & alerting and agent 9.4 -- audit & compliance.
framework-scout
Open-source agent framework assessor that evaluates community health, supersession risk, architecture alignment, and governance sustainability to classify investment readiness before committing resources.
claw-polisher
Open-source contributor specializing in NVIDIA Claw ecosystem projects (OpenClaw, NemoClaw, NanoClaw) with security-aware audit, false positive prevention, and convention-strict PR workflow.
agent-boundaries
Many OTAIP stages contain sibling agents with adjacent responsibilities. This guide states which agent owns which responsibility — the intent, field, or operation that assigns a request to exactly one agent — derived from the agent contracts.