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 skills add felipefontoura/bento --skill paperclipgit clone --depth 1 https://github.com/felipefontoura/bentoWrote 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/skills/felipefontoura/bento/paperclip)<a href="https://agentmods.dev/skills/felipefontoura/bento/paperclip"><img src="https://agentmods.dev/badge/skills/felipefontoura/bento/paperclip/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/felipefontoura/bento/paperclip"><img src="https://agentmods.dev/badge/skills/felipefontoura/bento/paperclip.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00131 | $0.02842 |
| Opus 5 | $0.00066 | $0.01421 |
| Sonnet 5 | $0.00026 | $0.00568 |
| Haiku 4.5 | $0.00013 | $0.00284 |
Grade A, and why
paperclip 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 9d 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 -s -H "authorization: Bearer $TOK" "https://$PAPERCLIP_HOST/api/companies/$COMPANY/agents" How it starts
The opening of the file, as written. The whole thing — 202 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You operate the application API of a paperclip stack that is already
deployed by bento. This is day-2 work: you configure agents and skills, manage
the board, and wire integrations. You do NOT install or redeploy the stack —
that is /bento:deploy. All artifacts stay in English.
Golden rule — mutate via the API, never the DB or disk. Creating agents, editing instruction bundles, syncing skills, and changing config must go through the HTTP API. The API validates schemas, applies defaults, fires the runtime side-effects the app expects (instruction-folder scaffold, config revisions, audit log, skill-content sync), and never leaves an agent half-built. Raw
INSERT/UPDATEon Postgres orcat > fileinto the container can diverge from what the app expects. The only tolerated DB exception is minting a bootstrap board key (below) — and you delete it after. Read-only DB/code inspection to understand the API is always fine.
When to invoke
- "create / configure a paperclip agent (CEO, chiefs, ICs)"
- "write the instruction bundle (SOUL / AGENTS / HEARTBEAT / TOOLS)"
- "import skills into paperclip" / "sync an agent's skills"
- "delete issues / clean up the board / reset failed runs"
- "wire the paperclip MCP server into hermes (control plane)"
For getting paperclip running use /bento:deploy. For provider API keys
(OpenAI/Anthropic) use /bento:auth — paperclip reads those propagated envs.
Discover the instance — don't hardcode
Read the host from bento state on the VPS (same source the other skills use):
ssh "$user@$host" "jq -r '.envs.paperclip.PAPERCLIP_HOST' \$HOME/.config/bento/state.json"
Base URL is https://<PAPERCLIP_HOST>. Find the company id (you almost always
need it — most routes are /api/companies/:companyId/...):
# from inside the box, against the loopback API (see headless mode below)
docker exec "$(paperclip_container)" node -e \
'fetch("http://127.0.0.1:3100/api/companies",{headers:{authorization:"Bearer "+process.env.TOK}}).then(r=>r.json()).then(d=>console.log(JSON.stringify(d,null,2)))'
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.
- 9d ago First seen · 202 lines · 131 tokens per session scan A 70b85796fd0e
paperclip is a skill published in the GitHub repository felipefontoura/bento (20 stars, last pushed 1mo ago), licensed MIT. It adds 131 tokens to every session and 2,842 once invoked, about $0.0007 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 skills, from other repositories
persona-event-coordinator
Plan and manage events — scheduling, invitations, and logistics.
persona-project-manager
Coordinate projects — track tasks, schedule meetings, and share docs.
persona-team-lead
Lead a team — run standups, coordinate tasks, and communicate.
recipe-create-meet-space
Create a Google Meet meeting space and share the join link.
recipe-plan-weekly-schedule
Review your Google Calendar week, identify gaps, and add events to fill them.
recipe-review-overdue-tasks
Find Google Tasks that are past due and need attention.