agentleFS
NVIDIAKnown publisherNVIDIA/nemoclaw-community/examples/recipes/partners/bluetier/x402-payment-gate/agents/hermes/skills/blackwall-payment-gate/SKILL.md

blackwall-payment-gate

Skill142 starsChanged 3 months agoApache-2.0View on GitHub

Screen x402 payments with an advisory Blackwall verdict, then submit a payment intent to the host-side release gate. You prepare payments; only the gate can settle them.

  • Sends data out

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Never attempt to reach a facilitator, payment rail, or wallet directly — the policy denies those routes, and an attempt is treated as a …

    line 56
  2. Very specific:

    Never resubmit a refused intent, and never split or resize a payment to turn a HOLD into a release.

    line 59
  3. Distinctive:

    Do not send anything except the claim fields (counterparty, amount, asset, chain, resource) to the verdict service or the gate.

    line 63

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

blackwall-payment-gate

You are the MAKER in a maker/checker payment boundary. You can screen payments, explain verdicts, and submit payment intents. You cannot sign or settle anything: this sandbox has no signing key and no network route to any payment rail or facilitator — that is a platform property, not a rule you could choose to break.

When to use

  • A tool call or resource fetch returned HTTP 402 with an x402 challenge and the user wants to pay it.
  • The user asks whether a counterparty is safe to pay, or why a payment was held or refused.

Procedure

  1. Extract from the x402 challenge: the payTo address (counterparty), the quoted amount, asset, chain, and the resource URL.
  2. Advisory pre-check (optional but preferred — it lets you warn the user before anything is submitted):

    python3 scripts/blackwall_client.py \
      --counterparty <payTo> --amount <amount> --resource <resource-url>

    GO means likely to release; HOLD/STOP mean expect the gate to hold or refuse — tell the user the reasons now.

  3. Submit the intent to the release gate (the only path to settlement):

    curl -sS -X POST http://host.openshell.internal:8790/v1/intents \
      -H 'Content-Type: application/json' \
      -d '{"counterparty":"<payTo>","amount":"<amount>","resource":"<url>"}'
  4. Report the gate's decision to the user, with the verdict reasons:
    • released — the gate's mandatory verdict was GO; it signed and settled.
    • held — escalated. A named human operator with the host-side approval token can release it; you cannot. Approval re-screens with a fresh verdict, so a payee that became sanctioned since submission is still refused. Give the user the intent id and the reasons.
    • refused — a hard signal fired (e.g. sanctions). Do not resubmit; explain which reason caused it.
  5. To answer later "what happened to that payment?" questions: curl -sS http://host.openshell.internal:8790/v1/intents/<id>

Rules

  • Never attempt to reach a facilitator, payment rail, or wallet directly — the policy denies those routes, and an attempt is treated as a boundary test, not a payment.
  • Never resubmit a refused intent, and never split or resize a payment to turn a HOLD into a release; escalate to the human operator instead.
  • The advisory pre-check and the gate use the same verdict service; a GO in step 2 is not a promise — the gate re-checks at release time.
  • Do not send anything except the claim fields (counterparty, amount, asset, chain, resource) to the verdict service or the gate.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.