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_invitegit clone --depth 1 https://github.com/magnusfroste/flowwinkWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/magnusfroste/flowwink/agent_invite)<a href="https://agentmods.dev/agents/magnusfroste/flowwink/agent_invite"><img src="https://agentmods.dev/badge/agents/magnusfroste/flowwink/agent_invite.svg" alt="Measured on agentmods" height="20"></a>What 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 | $0.00027 | $0.01524 |
| Opus 5 | $0.00014 | $0.00762 |
| Sonnet 5 | $0.00005 | $0.00305 |
| Haiku 4.5 | $0.00003 | $0.00152 |
Grade A, and why
agent_invite scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
curl -X POST "$INSTANCE_URL/functions/v1/mcp-server" \ How it starts
The opening of the file, as written. The whole thing — 142 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent invite payload
This is the template text an operator sends to an external agent (OpenClaw, Cursor, Claude Desktop, a custom client) when inviting it from Admin → Federation → Agents. Replace the two placeholders with the values from that page.
Never commit a real key.
<INSTANCE_URL>and<MCP_KEY>are per-instance secrets. An earlier revision of this file contained a live key; if you are auditing an instance, rotate any key that has ever been pasted into a repo, issue tracker or chat.
Connection
- Base URL:
<INSTANCE_URL>/functions/v1/mcp-server - Authentication:
Authorization: Bearer <MCP_KEY>
The server speaks two protocols on the same URL — both expose the same tools and resources. Pick whichever the client supports.
Option A — native MCP (JSON-RPC over Streamable HTTP)
For MCP-capable clients (Cursor, Claude Desktop, mcp-inspector, OpenClaw).
- Method:
POST /(root path) - Headers:
Authorization: Bearer <MCP_KEY>Content-Type: application/jsonAccept: application/json, text/event-stream— mandatory per the MCP Streamable HTTP spec; without it the server answers406.
- Methods:
initialize,tools/list,tools/call,resources/list,resources/read,notifications/*. - Surface control (avoids context bloat):
?groups=crm,commerce,hr— only those categories' tools. Live list atGET /rest/groups.?mode=dispatch— a 2-tool surface:search_skillsthenexecute_skill.- no parameter — every tool of every active module.
curl -X POST "$INSTANCE_URL/functions/v1/mcp-server" \
-H "Authorization: Bearer $MCP_KEY" \
-H "Content-Type: application/json" \
-H "Accept: application/json, text/event-stream" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"acquire_lock","arguments":{"lane":"lead_abc","ttl_seconds":60}}}'
Two failure modes worth knowing
| Symptom | Cause |
|---|---|
406 Not Acceptable |
Missing Accept: application/json, text/event-stream. |
Method not found (-32601) right after fixing the header |
The tool name doesn't exist. Discover real names with tools/list — e.g. there is no site_health_check; read the flowwink://health resource instead. |
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.
- 4d ago First seen · 142 lines · 27 tokens per session scan A c7a5328b5ccd
agent_invite is an agent published in the GitHub repository magnusfroste/flowwink (22 stars, last pushed today), licensed MIT. It adds 27 tokens to every session and 1,524 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
mcp-integration
Use this agent when any work touches packages/next/src/mcp/ or packages/next/src/tasks/, or more broadly when a change involves the MCP server, server-initiated notifications (e.g. taskassigned), the task store, the reporttaskdone / reporttaskprogress tools, or the UI-to-agent protocol. This agent MUST BE USED…
runtime-dev
MUST BE USED for any change inside packages/next/ in the agntcms monorepo, including the domain model, storage adapters, runtime, React components, runtime-level MCP glue, and the public subpath exports of @agntcms/next (/server, /client, /handlers, /config). This agent owns the @agntcms/next package. Note: deep MCP…
skills-dev
MUST BE USED for any change inside packages/skills/ or to the template/ directory layout, including folder structure, frozen files (app/api/agntcms/, app/[[...slug]]/page.tsx, .claude/, docker-compose.yml, Dockerfile), or the skill modules themselves. This agent owns @agntcms/skills and the canonical template…
cli-dev
Use this agent when any change is required inside packages/cli/ or when working on the onboarding flow for new agntcms projects. This agent owns the create-agntcms-app package and the create-agntcms-app command. MUST BE USED for any modification to CLI code, scaffolding logic, dependency installation steps…
barnabas
Use this agent for Discord server administration, community management, announcements, moderation, and member engagement. Specializes in the Cyber Defense Tactics Discord community. Reports to Chief Marketing Officer.
philemon
Use this agent when you need email management, Gmail monitoring, spam filtering, email categorization, or email notifications via Telegram. Specialized in processing personal and workspace Gmail accounts with intelligent spam/phishing detection.