# Hermes Plant

> Agent Action Safety - gate consequential shell, Git, SQL, deploy, x402, and MCP actions before execution - https://hermesplant.com

## Best short answer

Hermes Plant is an Agent Action Safety layer. Call the $0.01 quick gate before a consequential action; low risk returns to local policy for a decision, medium risk gets full DestructGuard evidence, and high or critical risk routes to the $0.25 complete workflow for review triage, a signed receipt, and a free status record. Human approval is never implied by triage. Assurance, validation, and finance utilities remain available.

## Who it is for

- Teams operating autonomous purchasing agents over x402 or MCP.
- Security, finance, and operations owners who need enforceable spend policy and signed evidence.
- Developers who need deterministic finance, validation, or safety tools inside agent workflows.

## Agent Action Safety (primary product)

- POST /api/agent-services/action-safety/quick - $0.01 deterministic preflight with risk, matched rule ids, and an exact conditional next call.
- POST /api/agent-services/action-safety/run - $0.25 complete score, conditional ReviewQueue triage, signed receipt, and 30-day privacy-preserving status record.
- GET /api/agent-workflows/action-safety/status/<workflowId> - free status lookup; new records store a versioned canonical action hash, audit ids, decision, and receipt metadata rather than the raw command; legacy records report binding unavailable.
- Agent skill: /.well-known/agent-skills/protect-agent-action/SKILL.md.
- Skill install page: /skills/action-safety.
- Install: `npx skills@1.5.18 add JesseGdotIO/hermesplant-mcp-server --skill protect-agent-action`.
- TypeScript SDK: `withHermesSafety` executes low risk, obtains full evidence for medium risk, and fails closed without executing high or critical actions.
- Access: 250 free calls/month with a free key, pay-per-call over x402, or the $29 pass with $35 of public-price-weighted credits.
- Boundary: review triage is not a human approval decision; the caller retains execution.

## Agent Commerce Assurance (hosted and pay-per-call trust rail)

- POST /api/assurance/decisions - evaluate an exact x402 or MCP action and receive a signed allow, deny, or review decision.
- POST /api/assurance/receipts - bind execution evidence to the exact prior decision.
- GET /api/assurance/history - query organization-scoped decision and receipt history.
- POST /api/agent-services/assurance/attest - pay-per-call ($0.05 over x402, no account): bind one x402 payment or MCP tool call (optional policy + settlement outcome) into a canonical HMAC-signed attestation record.
- POST /api/agent-services/assurance/verify - free: verify any Hermes-signed assurance record; recomputes the canonical hash and HMAC signature server-side.
- Every paid agent-service response also carries a `receipt` field (endpoint + SHA-256 of the exact response body, HMAC-signed) checkable at the same free verify endpoint.
- Authentication: signed-in organization session or organization bearer token for the organization APIs; pay-per-call attest needs no account (x402 or Agent API Pass) and verify is unauthenticated. Assurance plans (Team/Business) add organization history; legacy paid customers are grandfathered.
- Boundaries: Hermes Plant does not sign or execute payments; active probes are restricted to supported allowlisted checks.

## Enrichment & validation APIs (loop-callable — start here)

Deterministic per-record validators an agent calls once per record inside a loop: validate and normalize a contact list, screen every counterparty wallet. No data feed, no API key required; the same record always gets the same answer. Free tier: 250 calls/mo with a free API key (no wallet needed).

- [EmailGuard](https://hermesplant.com/agent-services/emailguard) - $0.02/call: Validate & normalize every contact record your agent touches.
- [WalletGuard](https://hermesplant.com/agent-services/walletguard) - $0.10/call: Review wallet risk signals you supply.

## Finance APIs

No API key and no data feed are required for the public paid endpoints. The caller supplies inputs, receives an x402 challenge, pays per call, and gets deterministic output with evidence.

- [CashflowLens](https://hermesplant.com/agent-services/cashflowlens) - $0.20/call: NPV, IRR, XIRR & DCF valuation in a single call.
- [WaterfallLens](https://hermesplant.com/agent-services/waterfall) - $0.25/call: LP/GP distribution waterfalls, solved exactly.
- [OptionLens](https://hermesplant.com/agent-services/options) - $0.25/call: Black-Scholes option pricing with the full Greeks.
- [BondLens](https://hermesplant.com/agent-services/bond) - $0.25/call: Yield, duration, convexity & loan amortization.
- [PortfolioGuard](https://hermesplant.com/agent-services/portfolioguard) - $0.15/call: Portfolio risk scored straight from holdings.
- [DealAnalyzer](https://hermesplant.com/agent-services/dealanalyzer) - $0.25/call: Full deal underwriting — DCF, returns, sensitivity & waterfall in one call.

## Agent trust and safety APIs (secondary)

- [Agent Action Safety Quick Gate](https://hermesplant.com/agent-services/action-safety) - $0.01/call: A one-cent safety gate for every consequential agent action.
- [Agent Action Safety Loop](https://hermesplant.com/agent-services/action-safety-workflow) - $0.25/call: Score, triage, receipt, and status in one agent-safety workflow.
- [x402 Demand Provenance Check](https://hermesplant.com/agent-services/x402-provenance) - $0.05/call: Separate candidate demand from owner-funded and duplicate x402 activity.
- [Assurance Attest](https://hermesplant.com/agent-services/assurance) - $0.05/call: A signed, verifiable receipt for every action your agent takes.
- [Payment Policy Decision API](https://hermesplant.com/agent-services/payment-policy) - $0.05/call: Decide whether an agent should pay before it signs.
- [Evidence Verification API](https://hermesplant.com/agent-services/evidence) - $0.05/call: Verify the evidence bundle before trusting a paid agent result.
- [MCP Server Risk Analyzer](https://hermesplant.com/agent-services/mcp-risk) - $0.05/call: Score an MCP server before your agent installs it.
- [DestructGuard Command Score](https://hermesplant.com/agent-services/destructguard) - $0.10/call: Catch destructive agent commands before they run.
- [ReviewQueue Agent Submit](https://hermesplant.com/agent-services/reviewqueue) - $0.25/call: Route risky agent actions to human approval.

## How an agent starts

1. Read the endpoint page or OpenAPI schema.
2. POST caller-owned inputs to the endpoint.
3. Inspect the 402 Payment Required challenge for method, path, amount, network, asset, payTo, facilitator, and resource binding.
4. Sign the x402 payment and retry the request.
5. Persist the response values, findings, and payment metadata in the agent run log.

## Endpoint reference pages

- [Agent Action Safety Quick Gate](https://hermesplant.com/agent-services/action-safety) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/action-safety/quick
- [Agent Action Safety Loop](https://hermesplant.com/agent-services/action-safety-workflow) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/action-safety/run
- [x402 Demand Provenance Check](https://hermesplant.com/agent-services/x402-provenance) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/x402-provenance/check
- [Assurance Attest](https://hermesplant.com/agent-services/assurance) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/assurance/attest
- [Payment Policy Decision API](https://hermesplant.com/agent-services/payment-policy) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/payment-policy/decide
- [Evidence Verification API](https://hermesplant.com/agent-services/evidence) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/evidence/verify
- [MCP Server Risk Analyzer](https://hermesplant.com/agent-services/mcp-risk) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/mcp-risk/score
- [DestructGuard Command Score](https://hermesplant.com/agent-services/destructguard) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/destructguard/score
- [ReviewQueue Agent Submit](https://hermesplant.com/agent-services/reviewqueue) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/reviewqueue/submit
- [EmailGuard](https://hermesplant.com/agent-services/emailguard) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/emailguard/validate
- [WalletGuard](https://hermesplant.com/agent-services/walletguard) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/walletguard/score
- [CashflowLens](https://hermesplant.com/agent-services/cashflowlens) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/cashflowlens/analyze
- [WaterfallLens](https://hermesplant.com/agent-services/waterfall) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/waterfall/distribute
- [OptionLens](https://hermesplant.com/agent-services/options) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/options/price
- [BondLens](https://hermesplant.com/agent-services/bond) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/bond/analyze
- [PortfolioGuard](https://hermesplant.com/agent-services/portfolioguard) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/portfolioguard/score
- [DealAnalyzer](https://hermesplant.com/agent-services/dealanalyzer) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/dealanalyzer/analyze
- [Agent Spend Assurance](https://hermesplant.com/agent-services/spend-assurance) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/spend-assurance/run
- [Investment Evidence Bundle](https://hermesplant.com/agent-services/investment-evidence) - endpoint docs, example request/response, input schema, and x402 call snippets. API path: /api/agent-services/investment-evidence/analyze

## Pricing and policy

- Pricing plans (free utility tier, Team Assurance $199/mo, Business Assurance $799/mo, scoped enterprise terms): https://hermesplant.com/pricing
- Live payment proof and evidence boundaries: https://hermesplant.com/proof
- API errors, status codes, retry semantics, rate limits: https://hermesplant.com/docs/api/errors
- Changelog (RSS): https://hermesplant.com/changelog/feed.xml
- Changelog (JSON Feed): https://hermesplant.com/changelog/feed.json
- Changelog (HTML): https://hermesplant.com/changelog
- Team and Business assurance are founder-assisted while production controls are validated with design partners. SSO, private gateways, custom exports, and contractual SLA terms are scoped separately.

## Machine-readable discovery

- Homepage: https://hermesplant.com
- API catalog: https://hermesplant.com/.well-known/api-catalog
- OpenAPI: https://hermesplant.com/openapi.json
- x402 manifest: https://hermesplant.com/.well-known/x402
- API docs: https://hermesplant.com/docs/api
- Auth policy: https://hermesplant.com/auth.md
- MCP server card: https://hermesplant.com/.well-known/mcp/server-card.json
- MCP endpoint: https://hermesplant.com/mcp
- Content API: https://hermesplant.com/api/content
- Feed: https://hermesplant.com/feed.json
- Agent skills: https://hermesplant.com/.well-known/agent-skills/index.json
- Example prompts (llms-prompts.txt): https://hermesplant.com/llms-prompts.txt
- Blog (founder-voice technical posts): https://hermesplant.com/blog

## Learn x402

- [What is x402? Explainer + glossary](https://hermesplant.com/guides/what-is-x402)
- [x402 and MCP: paid tools for AI agents](https://hermesplant.com/guides/x402-and-mcp-for-ai-agents)
- [x402 vs API keys](https://hermesplant.com/compare/x402-vs-api-keys)
- [x402 vs Stripe metered billing](https://hermesplant.com/compare/x402-vs-stripe-metered-billing)
- [x402 pay-per-call vs API subscriptions](https://hermesplant.com/compare/x402-pay-per-call-vs-api-subscriptions)

## Products (downloadable operator packs)

- [DestructGuard](https://hermesplant.com/products/destructguard): Free hosted blocklist rules editor and JSONL audit log viewer that stop AI coding agents from running destructive shell, git, SQL, and infra commands. No signup, no install.
- [IncidentScribe](https://hermesplant.com/products/incidentscribe): Free CLI and web tool that parse Cursor and Claude Code session JSONL into chronological incident timelines, flagging destructive shell commands, force pushes, and DB drops.
- [ReviewQueue](https://hermesplant.com/products/reviewqueue): Human-in-the-loop approval queue for risky AI-agent commands - review, approve, or deny from a web queue.
- [DestructGuard Pro Pack](https://hermesplant.com/products/destructguard-pro): Stop AI coding agents from running destructive commands. Full CLI with git hooks and audit log, plus 3 curated rules tiers and a team rollout playbook.
- [IncidentScribe Postmortem Pack](https://hermesplant.com/products/incidentscribe-postmortem-pack): Postmortem templates, team guide, and GitHub Action for agent session forensics.
- [ReviewQueue Team Pack](https://hermesplant.com/products/reviewqueue-team-pack): Slack/Telegram notifications, shared-queue guide, and audit rollup for agent approval teams.
- [MCP Risk Analyzer Operator Pack](https://hermesplant.com/products/mcp-risk-analyzer): Evidence-backed MCP server risk analyzer - scores every manifest on signing, permissions, code execution, network surface, and supply-chain risk. Pay-per-call x402 API for agents.
- [Operator OS - Revenue Solopreneur Core](https://hermesplant.com/products/operator-os-core): The 30-minute weekly system that runs your revenue: 8 AI prompts plus a Notion dashboard for solopreneurs.
- [Agent API Pass](https://hermesplant.com/products/agent-api-pass): $29/month with $35 of public-price-weighted endpoint credits, automatic per-call x402 fallback, and no raw-call arbitrage.
- [Assurance Team](https://hermesplant.com/products/assurance-team): Founder-assisted agent-commerce controls for one team: shared payment policy, signed decisions, receipt history, drift evidence, and rollout support.
- [Assurance Business](https://hermesplant.com/products/assurance-business): Founder-assisted assurance for production agent-commerce programs: organization controls, extended history, audit exports, SSO/security planning, and SLA design.

## Key guides

- [How AI agents pay for APIs with x402](https://hermesplant.com/guides/how-ai-agents-pay-for-apis-with-x402)
- [DCF and IRR for AI agents](https://hermesplant.com/guides/dcf-and-irr-for-ai-agents)
- [Deal underwriting for AI agents](https://hermesplant.com/guides/deal-underwriting-for-ai-agents-with-dealanalyzer)
- [Portfolio risk monitoring for AI agents](https://hermesplant.com/guides/portfolio-risk-monitoring-for-ai-agents)
- [FAQ](https://hermesplant.com/faq)

## Blog

- [How we ship x402 endpoints without confusing discovery with demand](https://hermesplant.com/blog/x402-discovery-without-fake-demand): An evidence-led release checklist for x402 catalogs, Coinbase Bazaar, APINow, agent skills, MCP directories, indexing, and real buyer-demand gates.
- [When to move from pay-per-call to Agent Ops](https://hermesplant.com/blog/agent-ops-vs-pay-per-call): A practical threshold model for deciding when AI agent traffic should stay on x402 pay-per-call versus switching to Agent Ops subscription plans.
- [LLM math failure modes in deal workflows](https://hermesplant.com/blog/llm-math-failure-modes-in-deal-workflows): Five recurring numerical failure modes we see in agent-led finance workflows, and how deterministic endpoint calls prevent expensive downstream errors.
- [MCP payments with policy guardrails in one loop](https://hermesplant.com/blog/mcp-payments-with-policy-guardrails): How to combine MCP tool calls, x402 settlement, and payment-policy checks so autonomous agents can buy API calls without blowing risk budgets.
- [x402 from curl: the smallest possible agent transaction](https://hermesplant.com/blog/x402-from-curl): What a 402 Payment Required response actually looks like, how an agent signs the EIP-712 payment authorization, and what comes back. Two curl commands and one signed authorization — that's the whole flow.

## Command safety hubs

- [Command safety hubs](https://hermesplant.com/commands)
- [Agent tool safety hubs](https://hermesplant.com/tools)
- [Explore by risk](https://hermesplant.com/explore)
- [aws s3 rm --recursive: what it deletes, safe flags, and recovery](https://hermesplant.com/commands/aws-s3-rm)
- [curl | sh: what it executes, why it's risky, and how to vet scripts](https://hermesplant.com/commands/curl-pipe-sh)
- [docker rm -f — why AI agents must not run it unattended](https://hermesplant.com/commands/docker-rm-force)
- [DROP DATABASE — why AI agents must not run it unattended](https://hermesplant.com/commands/drop-database)
- [git clean -fdx — why AI agents must not run it unattended](https://hermesplant.com/commands/git-clean-fdx)
- [What does git push --force do? Risks, --force-with-lease, recovery](https://hermesplant.com/commands/git-push-force)
- [git reset --hard explained: HEAD~1, reset to remote, and undo](https://hermesplant.com/commands/git-reset-hard)
- [kubectl delete — why AI agents must not run it unattended](https://hermesplant.com/commands/kubectl-delete)

## Founder / authority

- Operated in Jesse Gabriellini's project network. Founder hub: https://jesseg.io/portfolio/hermes-plant
- Hub landing: https://jesseg.io/hermes-plant

## Related founder projects

- SkillFrame (https://skillframe.co/)
- Revenu (https://revenuapp.com/)
- Phoenix Digital Growth (https://www.phoenixdigitalgrowthagency.com/)
- Phoenix Home Cleaning (https://phxhomecleaning.com/)

Support: contact@hermesplant.com