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 SELAT-AI/selat-plugins --skill selat-discoverygit clone --depth 1 https://github.com/SELAT-AI/selat-pluginsWrote 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/selat-ai/selat-plugins/selat-discovery)<a href="https://agentmods.dev/skills/selat-ai/selat-plugins/selat-discovery"><img src="https://agentmods.dev/badge/skills/selat-ai/selat-plugins/selat-discovery/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/selat-ai/selat-plugins/selat-discovery"><img src="https://agentmods.dev/badge/skills/selat-ai/selat-plugins/selat-discovery.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 | $0.00203 | $0.04627 |
| Opus 5 | $0.00102 | $0.02313 |
| Sonnet 5 | $0.00041 | $0.00925 |
| Haiku 4.5 | $0.00020 | $0.00463 |
Grade A, and why
selat-discovery scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
curl -s "https://api.apify.com/v2/store?search=instagram&limit=5" How it starts
The opening of the file, as written. The whole thing — 276 lines — stays where its author put it; the contents beside it link to each section on GitHub.
SELAT — discover, then pay (self-custody)
This is a concise DRIVER written for the plugin. The published @selat-ai/selat-discovery SKILL.md (bundled inside @selat-ai/selat-cli) remains the source of truth for exact subcommand flags, output shapes, and any commands added after the pinned CLI version. Where they conflict, the published skill wins. The command surface below was verified against @selat-ai/[email protected] and @selat-ai/[email protected].
SELAT is a capability layer for AI agents. It does two things a flat capability index doesn't: it checks vetted skills first, and it pays from the user's own wallet — SELAT never sees a private key or holds the user's funds.
The runner: how to invoke selat
All SELAT actions go through the selat runner (the published @selat-ai/selat-cli,
which bundles the discovery skill and the selat-pay engine). Resolve it in this order:
- Plain
selat— the SessionStart hook puts it on PATH (immediately on hosts that persist hook env; in new shells elsewhere). - If bare
selatdoesn't resolve, use the$SELAT_RUNNERenv var (an absolute path to the shim), where the host exports it. - If neither resolves, SELAT isn't available in this environment — tell the user rather than improvising. Do not hand-roll x402 requests or paste private keys.
Verify with selat doctor (read-only diagnostics: skill, PATH, Circle auth, Agent
Wallet, config).
Setup is the user's, not yours (self-custody)
Payments settle from the user's own Circle Agent Wallet (Circle MPC). SELAT never holds keys or balance. So:
- Free discovery needs no wallet —
selat search,selat skill list, andselat doctorrun with zero setup. Lead with those; show the user what's available first. - Only when the user wants to actually call/pay for a result and
selat doctorreports the wallet/config is missing: ask the user and wait beforeselat init, a Circle OTP, or any spend. Do not auto-runselat init. After they agree, it installs the Circle CLI, then prompts for the user's Circle email + a 6-digit OTP (relay those — the user authenticates; in a chat-only harness the OTP necessarily passes through the chat — keep it out of logs and summaries, and never store it). On a non-interactive host (no usable TTY) or an account with several agent wallets, passselat init --email <addr> --wallet <n|address>so it doesn't block on the email/wallet prompts —--wallettakes a 1-based index from the list init prints, a 0x address, ornew(also settable viaSELAT_WALLET); without it a multi-wallet account stops atWallet to use [1-N/new], which a non-TTY shell can't answer. Don't runselat initbefore it's needed, pre-install Circle CLI, or improvisecirclecommands.selat fundand any paid call still need explicit approval and an armed session budget (selat budget start);selat freezeis the kill switch — never auto-fund or auto-pay. - A
--raw-keydev mode exists but is not for production — do not steer users to it. - Before any spend, surface the cost and get the user's go-ahead. Spending limits are
set via
selat setup-policy(recommended before deposits > $20); funding viaselat fund. Both are user-driven money actions — never run them unprompted. - Funding is two movements: (1) top up the agent wallet address with USDC, then (2)
selat funddeposits it into Gateway. For movement (1) there are two paths: a crypto transfer to the wallet address, or fiat viaselat fund --onramp [--address 0x…](selat-cli 0.15.17+) — it mints a Circle Onramp browser link, chain-scoped so purchases can't strand. The user picks the amount inside the widget — never ask for an amount before launching, and never treat the mint as a spend (nothing moves until the user completes the purchase in their browser; relay the printed URL into chat). To onramp a wallet other than the configured payer, pass--address 0x…— never rewrite the configured wallet to retarget funding. Never usecircle wallet fund --method fiatfor fiat: that opens Transak, a different, unscoped onramp. After funding, refer only to "the Gateway balance" — one number, spendable on any supported chain; never enumerate per-chain balances to the user. Verify it withcircle gateway balance --all, or the Gateway line inselat doctor— never the wallet's on-chain address balance. A deposit moves USDC from the address into Gateway, so the on-chain balance dropping (even to 0) is by design, not lost money. Deposits take ~5–10 min to settle; a fresh 0 Gateway reading usually means "still settling" — don't tell the user funds are lost. Ifcircle wallet balanceand an on-chain read disagree, say "a fresh transfer can take a minute to be readable — let's re-check" — never speculate that Circle "holds" funds off-chain (reads just lag; that speculation manufactures a custodial misconception). Whenselat fundprints a QR image path, send that image to the user directly in chat harnesses — never just relay a file path the user can't open.
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 · 276 lines · 203 tokens per session scan A 429efd8a71a4
selat-discovery is a skill published in the GitHub repository SELAT-AI/selat-plugins (1 stars, last pushed 2d ago), licensed Apache-2.0. It adds 203 tokens to every session and 4,627 once invoked, about $0.0010 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 2026-08-31.
Other skills, from other repositories
skill-creator
Use this skill when a contributor wants to build, author, scaffold, verify, or submit a new skill to the SELAT skill hub (selat-skills) — e.g. "create a skill", "build a selat skill", "add a skill to the hub", "contribute a skill", "how do I write a manifest.json", "verify my skill", "submit my skill", "wrap an MPP…
self-evolving-agent
Use this skill when the user wants to design or operate a budgeted economic agent with its own operational identity, AgentMail address, agent-wallet treasury, infrastructure budget, social and financial intelligence loop, monetization or trading hypotheses, and reinvestment policy. The skill enforces treasury…
perplexity-search
Use this skill when the user wants a grounded web answer or research from Perplexity without an API key — e.g. "search the web for ", "what's the latest on ", "pull cited web context on ", "research with sources", "do a deep-research report on ". Runs Perplexity's x402 endpoints (search / sonar answer / async…
twitter-research
Use this skill for read-only Twitter/X research on the SELAT-native Twitter API (catalog.selat.ai) — profiles, recent tweets, mentions, followers, tweet details/replies/retweeters, topic search, and trends. Triggers on "who is @X on Twitter", "recent tweets from X", "who's mentioning X", "how did this tweet do / who…
enrich-waterfall
Use this skill when the user wants to enrich a person or company from a partial identifier — e.g. "enrich this email", "find the work email and phone for X at Y", "who is this person", "build a lead profile for stripe.com", "company enrichment for this domain", "find their LinkedIn / Twitter", "get me funding and…
stock-direction-signals
Use this skill when the user wants a bullish, bearish, or mixed directional read on a US stock — e.g. "is NVDA bullish or bearish right now", "give me a directional read on AAPL", "is MAG7 sentiment turning", "what do the chart, news, and social say about AMD", "signal brief on SPY". Covers MAG7 names, semiconductor…