# Agent Trust Gate

Public repo: https://github.com/Gareth1953/agent-trust-gate

Agent Trust Gate is a local-first, pre-action and pre-settlement trust gate for
agent-led workflows. It demonstrates local checks for mandate, evidence,
verified intent, receipt/proof validity, replay protection, and settlement
blocking before any sensitive action would proceed.

Core line:
No mandate. No evidence. No verified intent. No signed gate pass. No settlement.

Strategic boundary:
Readable now. Callable later. Autonomous execution never without gate control.

Key trust principle:
Claimed agent identity is not trust. Behaviour, mandate, evidence, verified
intent, and session context must decide access.

P3-M155 agent discovery boundary:
Software agents and accountable operators may statically inspect ATG, run local
deterministic demonstrations, and prepare a synthetic or properly sanitised
scenario. An accountable human must decide whether it is shared and approve any
technical or commercial engagement. ATG is not a live A2A server, MCP server,
hosted API, remote agent, enrolment service, payment service, or autonomous
buyer/contact system.

Safe local commands:
- npm test
- npm run build
- npm run typecheck
- npm run cli -- help
- npm run gate
- npm run demo:adversarial
- npm run demo:integrations
- npm run demo:quickstart
- npm run demo:session-intent
- npm run demo:prove-yourself
- npm run demo:agent-proof-contract
- npm run demo:agent-proof-integration
- npm run demo:enforceable-tool-gate
- npm run demo:gatepass-core
- npm run demo:gatepass-round-trip
- npm run demo:agent-readable-benefit
- npm run demo:gatepass-trust-language
- npm run demo:gatepass-scorecard
- npm run demo:gatepass-wrapper
- npm run demo:reviewer-kit
- npm run demo:paid-pilot
- npm run demo:discovery
- npm run demo:commerce-gatepass
- npm run demo:agent-standing -- --summary-only
- npm run demo:agent-invitation
- npm run demo:agent-invitation -- --summary-only
- npm run demo:agent-invitation -- --json
- npm run validate:agent-invitation
- npm run validate:discovery-site
- npm run proof:money-gate

Important docs:
- README.md
- agent-trust-gate.discovery.json
- agent-trust-gate.agent-review-invitation.json
- schemas/agent-review-invitation.schema.json
- schemas/bring-your-agent-scenario.schema.json
- examples/bring-your-agent-scenario.example.json
- docs/agent-standing.md
- docs/agent-discovery-and-reviewer-invitation.md
- docs/bring-your-agent-scenario.md
- docs/agent-discovery-standards-boundary.md
- docs/agent-readable-discovery-and-system-metadata.md
- docs/system-integration-metadata.md
- docs/example-agent-discovery-prompts.md
- docs/schema-formalisation-and-evidence-model.md
- docs/local-signed-receipt-and-proof-prototype.md
- docs/adversarial-evaluation-pack.md
- docs/simplified-developer-cli.md
- docs/reference-integration-examples.md
- docs/paid-pilot-readiness-review.md
- docs/public-reviewer-and-paid-pilot-enquiry-pack.md
- docs/reviewer-enquiry-copy.md
- docs/paid-pilot-enquiry-checklist.md
- docs/paid-pilot-commercial-entry.md
- docs/paid-pilot-scope-and-deliverables.md
- docs/buyer-evaluation-journey.md
- docs/paid-pilot-pricing-boundary.md
- docs/paid-pilot-enquiry-template.md
- docs/embedded-commerce-gatepass.md
- docs/embedded-commerce-design-partner-pilot.md
- docs/machine-discovery-and-registry-readiness.md
- docs/machine-readable-entry-points.md
- docs/github-pages-discovery-readiness.md
- docs/a2a-discovery-readiness-boundary.md
- docs/mcp-registry-readiness-boundary.md
- docs/npm-publication-readiness.md
- docs/registry-readiness-scorecard.md
- docs/public-repo-commercial-conversion-review.md
- docs/buyer-use-cases-and-revenue-triggers.md
- docs/revenue-trigger-map.md
- docs/paid-review-scope-examples.md
- docs/global-code-discovery-and-developer-distribution-pack.md
- docs/github-discovery-metadata-guide.md
- docs/developer-distribution-checklist.md
- docs/global-developer-sharing-copy.md
- docs/agent-readable-distribution-note.md
- docs/controlled-public-visibility-and-paid-enquiry-positioning.md
- docs/public-visibility-readiness-checklist.md
- docs/paid-enquiry-positioning.md
- docs/public-positioning-message-bank.md
- docs/controlled-distribution-sequence.md
- docs/agent-trust-invitation-and-prove-yourself-protocol.md
- docs/agent-proof-requirements.md
- docs/system-side-agent-verification-guide.md
- docs/agent-owner-trust-presentation-guide.md
- docs/what-a-gate-pass-proves.md
- docs/agent-proof-package-schema-and-verification-contract.md
- docs/agent-proof-package-field-guide.md
- docs/gate-pass-challenge-and-response-flow.md
- docs/agent-proof-contract-integration-readiness.md
- docs/local-agent-workflow-integration-guide.md
- docs/tool-calling-proof-gate-adapter-guide.md
- docs/pre-settlement-proof-contract-integration.md
- docs/integration-readiness-checklist.md
- docs/enforceable-local-tool-calling-gate-demo.md
- docs/local-tool-call-gate-wrapper-guide.md
- docs/mock-sensitive-tools-catalog.md
- docs/tool-call-enforcement-scenarios.md
- docs/what-the-enforceable-tool-gate-demo-proves.md
- docs/minimal-gatepass-core-specification.md
- docs/gatepass-field-guide.md
- docs/gatepass-minimal-profile.md
- docs/gatepass-proofpackage-consolidation.md
- docs/why-minimal-gatepass-matters.md
- docs/gatepass-create-verify-reject-round-trip.md
- docs/gatepass-round-trip-developer-guide.md
- docs/gatepass-rejection-reason-catalog.md
- docs/gatepass-round-trip-threat-model.md
- docs/what-the-gatepass-round-trip-proves.md
- docs/agent-readable-gatepass-benefit-and-voluntary-referral.md
- docs/why-agents-benefit-from-gatepass.md
- docs/voluntary-gatepass-referral-rules.md
- docs/agent-readable-benefit-statement.md
- docs/safe-agent-recommendation-language.md
- docs/agent-trust-language-and-gatepass-vocabulary.md
- docs/gatepass-trust-language-vocabulary.md
- docs/agent-trust-language-phrasebook.md
- docs/agent-to-system-trust-dialogue-examples.md
- docs/gatepass-trust-language-safety-boundary.md
- docs/gatepass-adversarial-metrics-and-latency-scorecard.md
- docs/gatepass-metrics-methodology.md
- docs/gatepass-adversarial-scenario-catalog.md
- docs/gatepass-latency-measurement-guide.md
- docs/gatepass-reviewer-scorecard-guide.md
- docs/gatepass-developer-wrapper-and-local-integration-example.md
- docs/gatepass-wrap-tool-developer-guide.md
- docs/local-agent-framework-integration-example.md
- docs/gatepass-wrapper-policy-guide.md
- docs/gatepass-wrapper-limitations-and-safety-boundary.md
- docs/one-command-reviewer-demo-kit.md
- docs/reviewer-demo-kit-quickstart.md
- docs/reviewer-demo-output-guide.md
- docs/reviewer-demo-limitations-and-safety-boundary.md
- docs/reviewer-evaluation-checklist.md
- docs/ai-agent-traffic-and-session-intent-gate.md
- docs/spoofed-agent-risk-model.md
- docs/session-specific-access-framework.md

Important schemas:
- schemas/local-agent-action-request.schema.json
- schemas/local-trust-receipt.schema.json
- schemas/local-money-gate-proof.schema.json
- schemas/agent-proof-package.schema.json
- schemas/agent-proof-verification-request.schema.json
- schemas/agent-proof-verification-result.schema.json
- schemas/gate-pass-challenge.schema.json
- schemas/gatepass-core.schema.json

Important examples:
- examples/local-demo-low-risk-allow.json
- examples/local-receipt-signed-gate-pass.json
- examples/local-signed-trust-receipt-valid.json
- examples/local-signed-money-gate-proof-valid.json
- examples/adversarial-replay-blocked.json
- examples/adversarial-valid-control-allowed.json
- examples/reference-generic-agent-loop.json
- examples/reference-pre-settlement-money-gate.json
- examples/session-intent-spoofed-agent-blocked.json
- examples/session-intent-valid-local-control.json
- examples/prove-yourself-claimed-identity-only-blocked.json
- examples/prove-yourself-valid-local-control.json
- examples/agent-proof-package-valid-local-control.json
- examples/agent-proof-package-identity-only-invalid.json
- examples/agent-proof-verification-result-allowed-local.json
- examples/agent-proof-integration-valid-local-workflow.json
- examples/agent-proof-integration-sensitive-tool-call-escalated.json
- examples/agent-proof-integration-missing-proof-requires-evidence.json
- examples/agent-proof-integration-pre-settlement-requires-signed-proof.json
- examples/agent-proof-integration-high-risk-human-review.json
- examples/agent-proof-integration-replayed-proof-blocked.json
- examples/enforceable-tool-gate-public-post-allowed-local.json
- examples/enforceable-tool-gate-customer-message-escalated.json
- examples/enforceable-tool-gate-data-export-blocked.json
- examples/enforceable-tool-gate-prepare-payment-requires-signed-proof.json
- examples/enforceable-tool-gate-procurement-stale-proof-blocked.json
- examples/enforceable-tool-gate-high-risk-human-review.json
- examples/enforceable-tool-gate-settlement-instruction-blocked.json
- examples/enforceable-tool-gate-valid-local-control.json
- examples/enforceable-tool-gate-missing-proof-requires-evidence.json
- examples/gatepass-core-valid-local-low-risk.json
- examples/gatepass-core-sensitive-tool-valid.json
- examples/gatepass-core-high-risk-human-review.json
- examples/gatepass-core-pre-settlement-valid-local.json
- examples/gatepass-core-identity-only-invalid.json
- examples/gatepass-core-missing-mandate-invalid.json
- examples/gatepass-core-missing-evidence-invalid.json
- examples/gatepass-core-stale-expiry-invalid.json
- examples/gatepass-core-missing-signature-pre-settlement-invalid.json
- examples/gatepass-round-trip-valid-allow-local.json
- examples/gatepass-round-trip-identity-only-rejected.json
- examples/gatepass-round-trip-replayed-nonce-rejected.json
- examples/gatepass-round-trip-tampered-scope-rejected.json
- examples/gatepass-round-trip-pre-settlement-requires-signed-proof.json
- examples/gatepass-round-trip-pre-settlement-valid-local.json
- examples/agent-readable-gatepass-benefit-statement.json
- examples/gatepass-trust-language-vocabulary.json
- examples/gatepass-adversarial-scorecard.json
- examples/gatepass-developer-wrapper-demo.json
- examples/gatepass-reviewer-kit-report.json
- examples/paid-pilot-offer.json
- examples/commerce-valid-basket.json
- examples/commerce-unauthorised-substitution.json
- examples/commerce-price-cap-breach.json
- examples/commerce-address-change.json
- examples/commerce-expired-approval.json
- examples/commerce-replay-attempt.json
- examples/commerce-missing-evidence.json
- examples/embedded-commerce-gatepass-report.json
- examples/machine-discovery-report.json

Commercial/contact route:
Human-reviewed developer, agent-system, integration, security, paid evaluation
pilot, embedded-commerce design-partner scoping, paid technical review, local
pilot discussion, and public project enquiries may use
gpmiddleton71@gmail.com. This is contact metadata only and does not grant
automatic acceptance, automatic access, hosted access, payment activation,
payment links, checkout, live invoices, PayPal API integration, Stripe
integration, or settlement capability.

Safety boundary:
This repository is readable by agents and systems, but it exposes no live agent
endpoint and grants no autonomous authority. It has no live API access, no MCP
server functionality, no live agent-to-agent communication, no agent
negotiation, no autonomous contact, no external-agent contact, no live payment
processing, no PayPal API integration, no Stripe integration, no checkout, no
webhooks, no wallet/banking logic, no real settlement execution, no production
signing, no production key management, no real tool execution, no cloud/network calls, no AUC
integration, no Agent Contact System integration, no outreach automation, and
no action execution.

Machine discovery boundary:
P3-M142 adds passive machine-readable project metadata only. It creates no live
A2A server, no A2A operational endpoint, no MCP server, no MCP Registry
publication, no npm publication, no registry credential, no live API endpoint,
no payment integration, no real tool execution, no settlement execution, and
no action execution. P3-M143A records GitHub Pages passive discovery as active
and HTTPS verified for static metadata only.

Passive discovery activation boundary:
P3-M143 prepared GitHub Pages passive discovery activation. It added
`.github/workflows/deploy-discovery-pages.yml`, static site polish, local
validation, an activation runbook, a live verification checklist, a social
preview manual route, and an activation record template. P3-M143A records the
public passive discovery site as active and verified:
`https://gareth1953.github.io/agent-trust-gate/`.
It does not create a custom domain, generate an image, add
analytics/tracking/forms/payment links, create A2A or MCP services, publish
npm, push, tag, release, expose a live API, process payments, execute
settlement, or execute actions.

Embedded commerce boundary:
P3-M144 adds a local deterministic Embedded Commerce GatePass demonstrator for
synthetic pre-checkout basket verification. It compares final proposed baskets
against a buyer mandate, limits, substitution policy, delivery reference,
currency, approval state, basket hash, nonce, and evidence. It can emit a
local commerce GatePass or a refusal receipt.

P3-M144A positions Embedded Commerce GatePass as a featured commercial
application for supermarkets, grocery retailers, general retailers,
AI-shopping platforms, commerce infrastructure teams, payment and checkout
providers, retail-system architects, and AI governance or transaction-risk
teams. It remains a local synthetic demonstrator and paid
evaluation/design-partner opportunity only.

Commerce support line:
No verified basket. No valid mandate. No current approval. No GatePass. No
checkout.

Design-partner principle:
We prove the trust architecture. The design partner funds the real integration.

Commerce command:
`npm run demo:commerce-gatepass`

Commerce docs and examples:
- `docs/embedded-commerce-gatepass.md`
- `docs/embedded-commerce-design-partner-pilot.md`
- `examples/embedded-commerce-gatepass-report.json`

This is synthetic evaluation only. It adds no live retailer integration,
shopping-agent integration, checkout service, account login, card handling,
payment token handling, payment processing, settlement execution, AI-provider
integration, network calls, A2A server, MCP server, production signing, or
action execution.

## P3-M144A - Embedded Commerce Commercial Positioning

P3-M144A strengthens public discovery and documentation for Embedded Commerce
GatePass as a featured commercial application of Agent Trust Gate. ATG remains
a broad trust layer for AI-agent actions, delegated decisions, and
pre-settlement controls; it is not repositioned as a supermarket-only product.

Target audiences include supermarkets, grocery retailers, general retailers,
AI-shopping platforms, commerce infrastructure teams, payment and checkout
providers, retail-system architects, developers, system architects, AI
governance teams, and transaction-risk reviewers.

The commercial route is a paid evaluation or design-partner pilot. Possible
synthetic evaluation scope includes mandate enforcement, basket integrity,
substitutions, price and fee limits, approval freshness, merchant and
destination checks, replay protection, and GatePass or refusal-receipt outputs.

No named-company affiliation, live retailer deployment, live checkout control,
payment processing, API, network call, A2A service, MCP service, production
deployment, release, tag, publication, push, or action execution is introduced.

## P3-M141 - Paid Pilot and Commercial Entry Pack

P3-M141 adds the Agent Trust Gate Paid Evaluation Pilot as a human-reviewed,
local, manual-input only, non-production, non-custodial, non-autonomous,
advisory evaluation route. The reviewer kit remains the recommended first
experience; GatePass remains the headline proof primitive; Agent Trust
Language remains supporting material only. Indicative pricing starts from
£1,500 for a defined local evaluation pilot, subject to scope and written
agreement.

Paid pilot command:
`npm run demo:paid-pilot`

Paid pilot docs:
- `docs/paid-pilot-commercial-entry.md`
- `docs/paid-pilot-scope-and-deliverables.md`
- `docs/buyer-evaluation-journey.md`
- `docs/paid-pilot-pricing-boundary.md`
- `docs/paid-pilot-enquiry-template.md`
- `examples/paid-pilot-offer.json`

Paid pilot safety flags: localDemoOnly true, manualInputOnly true,
humanApproved true, nonProduction true, nonCustodial true, nonAutonomous true,
advisoryOnly true, realToolExecution false, actionExecution false,
networkCalls false, liveApi false, mcpServerFunctionality false,
paymentIntegration false, paymentLinks false, checkout false,
livePaymentProcessing false, paypalApiIntegration false, stripeIntegration
false, webhookIntegration false, walletBankingLogic false, settlementExecution
false, productionSigning false, productionCertification false,
securityCertification false, legalComplianceGuarantee false,
guaranteedResults false, automaticAcceptance false, and
automaticAccessAfterPayment false.

## P3-M142 - Machine Discovery and Registry Readiness Pack

P3-M142 adds a canonical passive machine-discovery record, local discovery CLI,
registry-readiness boundary docs, and static discovery-site source. It improves
how humans, search systems, AI assistants, developers, and reviewers understand
Agent Trust Gate without adding executable remote services.

Canonical discovery record:
`agent-trust-gate.discovery.json`

Discovery command:
`npm run demo:discovery`

Discovery docs:
- `docs/machine-discovery-and-registry-readiness.md`
- `docs/machine-readable-entry-points.md`
- `docs/github-pages-discovery-readiness.md`
- `docs/a2a-discovery-readiness-boundary.md`
- `docs/mcp-registry-readiness-boundary.md`
- `docs/npm-publication-readiness.md`
- `docs/registry-readiness-scorecard.md`
- `examples/machine-discovery-report.json`

Discovery safety flags: localDemoOnly true, manualInputOnly true,
humanApprovalRequired true, nonProduction true, a2aServer false, mcpServer
false, mcpRegistryPublication false, npmPublication false,
githubPagesDeployment true, githubPagesActive true, githubPagesHttpsVerified
true, operationalEndpoint false, registryCredentials false, realToolExecution
false, actionExecution false, networkCalls false, livePaymentProcessing false,
settlementExecution false, productionSigning false, productionCertification
false, securityCertification false, and legalComplianceGuarantee false.

The reviewer kit remains the recommended first experience. GatePass remains
the headline proof primitive. Agent Trust Language remains supporting material.
GitHub topics are documented as manually configured through GitHub and are not
altered automatically by this repository. GitHub Pages is active as a static
passive discovery route only. No A2A server, A2A operational endpoint, MCP
server, MCP Registry publication, npm publication, outreach, bot messaging,
payment integration, product network call, tag, release, publish, push, or
action execution is introduced.

## P3-M143 - Passive Discovery Activation

P3-M143 prepared the existing P3-M142 static discovery site for controlled
GitHub Pages activation. P3-M143A records that route as active and verified.
The reviewer kit remains the recommended first experience and GatePass remains
the headline product concept.

Live Pages URL:
`https://gareth1953.github.io/agent-trust-gate/`

Prepared workflow:
`.github/workflows/deploy-discovery-pages.yml`

Prepared static source:
- `discovery-site/index.html`
- `discovery-site/404.html`
- `discovery-site/robots.txt`
- `discovery-site/sitemap.xml`
- `discovery-site/.nojekyll`
- `discovery-site/README.md`

Activation docs:
- `docs/github-pages-passive-discovery-activation.md`
- `docs/passive-discovery-live-verification-checklist.md`
- `docs/repository-social-preview-upload.md`
- `docs/passive-discovery-activation-record.md`
- `docs/passive-discovery-activation-record-template.md`

Local validation command:
`npm run validate:discovery-site`

Machine-readable status:
- `githubPagesDeploymentWorkflowPrepared` true
- `githubPagesActive` true
- `githubPagesHttpsVerified` true
- `githubPagesDeployment` true
- `networkCalls` false
- `actionExecution` false

The Pages artifact is intended to include only `discovery-site/`,
`agent-trust-gate.discovery.json`, `agent-trust-gate.agent-card.json`,
`agent-trust-gate.manifest.json`, and `llms.txt`. No live A2A endpoint, MCP
server, npm publication, payment integration, analytics, tracking, form,
custom domain, generated image, live API, tag, release, push, or action
execution is introduced.

## P3-M144 - Embedded Commerce GatePass Demonstrator

P3-M144 adds a local deterministic Embedded Commerce GatePass demonstrator for
synthetic pre-checkout basket verification. It keeps `npm run
demo:reviewer-kit` as the recommended first experience and treats commerce as
an optional specialist scenario.

Commerce command:
`npm run demo:commerce-gatepass`

Commerce docs:
- `docs/embedded-commerce-gatepass.md`
- `docs/embedded-commerce-design-partner-pilot.md`
- `examples/embedded-commerce-gatepass-report.json`

The demo can emit either a local commerce GatePass or a structured refusal
receipt. It proves the trust architecture locally; any real integration would
require paid design-partner scope, security review, and written agreement.

No live retailer integration, shopping-agent integration, checkout service,
account login, card handling, payment token handling, payment processing,
settlement execution, AI-provider integration, network call, A2A server, MCP
server, npm publication, production signing, production-grade cryptography,
real customer data, or action execution is introduced.

Agent-readable GatePass benefit boundary:
Agents may explain GatePass only in relevant trust, proof, authorisation,
mandate, evidence, sensitive tool use, auditability, pre-settlement, or
verification contexts. Do not use this repo for autonomous marketing, direct
bot messaging, hidden viral distribution, unsolicited agent contact, scraping,
contact harvesting, guaranteed trust claims, or proven-safe claims about an
agent.

GatePass proof vocabulary / claims vocabulary boundary:
GatePass Trust Language remains supporting GatePass proof vocabulary and claims
vocabulary, not a safety guarantee. It uses scoped, action-specific,
time-bounded language and rejects proven-safe, guaranteed-trust,
bypass-verification, autonomous marketing, and viral promotion claims. It
grants no live systems contact, direct bot messaging, live agent-to-agent
communication, hidden viral distribution, or action execution.

GatePass scorecard boundary:
P3-M137 adds a local deterministic GatePass adversarial metrics and latency
scorecard. It uses local illustrative timing only. It is not a production
benchmark, not a security certification, not adversarial completeness, and not
evidence of production readiness. localDemoOnly is true; productionBenchmark,
securityCertification, productionCertification, adversarialCompleteness,
liveToolExecution, actionExecution, paymentAuthorisation,
settlementAuthorisation, and networkCalls remain false.

GatePass developer wrapper boundary:
P3-M138 adds a local deterministic GatePass developer wrapper and local
framework-style integration example. It shows wrapGatePassTool gating local
mock tool calls before action. localDemoOnly and mockToolExecutionOnly are
true; realToolExecution, actionExecution, networkCalls,
liveAgentFrameworkDependency, productionMiddleware, productionCertification,
securityCertification, paymentAuthorisation, and settlementAuthorisation
remain false. It is not production middleware, not a live framework
integration, and not real tool execution.

GatePass reviewer kit boundary:
P3-M139 adds a local deterministic one-command reviewer demo kit. It runs the
GatePass lifecycle summary, adversarial scorecard summary, developer wrapper
summary, local integration summary, safety boundary, and JSON report output.
localDemoOnly and mockToolExecutionOnly are true; realToolExecution,
actionExecution, networkCalls, productionMiddleware, productionBenchmark,
productionCertification, securityCertification, legalComplianceGuarantee,
paymentAuthorisation, settlementAuthorisation, and
liveAgentToAgentCommunication remain false. It is not production middleware,
not a production benchmark, not security certification, and not production
readiness.

P3-M123 adds a local-only AI agent traffic and session intent gate concept. It
is not a web bot detection product, does not monitor live website traffic, does
not classify real traffic, does not block crawlers or browsers, and does not
perform browser fingerprinting, scraping, tracking, analytics, or telemetry.

P3-M124 adds human-approved public reviewer and paid pilot enquiry guidance.
It is manual-use documentation and copy only. It does not add outreach
automation, scraping or contact harvesting, forms, tracking, analytics,
telemetry, live payment processing, settlement, live APIs, external-agent
contact, automatic acceptance, automatic access, production signing, or action
execution.

P3-M125 adds a public repo commercial conversion review and README reviewer
start path improvements. It is documentation and navigation clarity only. It
does not add forms, tracking, analytics, hosted services, outreach automation,
scraping or contact harvesting, payment processing, settlement, live APIs,
external-agent contact, production signing, or action execution.

P3-M126 adds buyer use cases, a revenue trigger map, and paid review scope
examples. It is commercial documentation and positioning only. It does not
claim guaranteed buyer demand, guaranteed revenue, guaranteed paid-pilot
conversion, production readiness, live payment/settlement readiness, automatic
acceptance, automatic access after payment, or any legal, financial,
compliance, procurement, settlement, or security assurance claim. It adds no outreach
automation, scraping or contact harvesting, forms, tracking, analytics, hosted
services, live APIs, payment processing, settlement, external-agent contact,
production signing, or action execution.

P3-M127 adds global code discovery guidance, GitHub metadata guidance, a manual
developer distribution checklist, manual sharing copy, and an agent-readable
distribution note. It is documentation and manual distribution planning only.
It does not claim guaranteed global discovery, guaranteed buyer demand,
guaranteed paid-pilot conversion, production readiness, live
payment/settlement readiness, automatic acceptance, automatic access after
payment, or any legal, financial, compliance, procurement, settlement, or
security assurance claim. It adds no email automation, outreach automation, scraping
or contact harvesting, forms, tracking, analytics, paid ads, ad pixels, hosted
services, live APIs, payment processing, settlement, external-agent contact,
production signing, or action execution.

P3-M128 adds controlled public visibility and paid enquiry positioning, a
public visibility readiness checklist, paid enquiry positioning, a public
message bank, and a controlled distribution sequence. It is documentation,
positioning, manual public-visibility planning, and paid enquiry clarity only.
It does not claim guaranteed public visibility, guaranteed global discovery,
guaranteed buyer demand, guaranteed paid pilot conversion, production
readiness, live payment/settlement readiness, automatic paid-pilot acceptance,
automatic access after payment, or any legal, financial, compliance,
procurement, settlement, or security assurance claim. It adds no outreach automation,
email automation, scraping or contact harvesting, forms, tracking, analytics,
paid ads, ad pixels, hosted services, live APIs, payment processing,
settlement, external-agent contact, production signing, or action execution.

P3-M129 adds an Agent Trust Invitation and Prove-Yourself Protocol Pack with
local proof requirements, system-side verification guidance, agent-owner
presentation guidance, gate-pass meaning, deterministic examples, and a local
demo command. It states: Do not trust the agent. Trust the gate pass. No proof.
No permission. No mandate. No action. No signed gate pass. No settlement. It
does not claim universal agent standard status, guaranteed agent adoption,
identity certification, production readiness, live payment/settlement
readiness, legal/compliance/security certification, automatic acceptance, or
automatic access after payment. It adds no live systems contact, direct bot
messaging, live agent-to-agent communication, live APIs, payment processing,
settlement, external-agent contact, production signing, or action execution.

P3-M130 adds an Agent Proof Package Schema and Verification Contract Pack with
local proof package schemas, verification request/result schemas, gate-pass
challenge schemas, deterministic local examples, and a local demo command. It
turns the prove-yourself protocol into machine-readable local proof contracts:
claimed identity alone is not sufficient proof, no proof means no permission,
stale or replayed proof fails, and settlement-sensitive requests require signed
proof. It adds no live API, MCP server functionality, live systems contact,
direct bot messaging, live agent-to-agent communication, payment processing,
settlement, hosted service, production signing, production certification, or
action execution.

P3-M131 adds an Agent Proof Contract Integration Readiness Pack with local
workflow, tool-call, approval, governance, session/access, and pre-settlement
adapter guidance, deterministic local examples, and a local demo command. It
keeps the proof contract local-only and adds no live API, MCP server
functionality, live systems contact, direct bot messaging, live agent-to-agent
communication, payment processing, settlement execution, hosted service,
production signing, production certification, or action execution.

P3-M132 adds an Enforceable Local Tool-Calling Gate Demo with runnable local
mock-agent tool-call interception, proof/policy checks, receipt-style results,
deterministic examples, and a local demo command. It adds no real tool
execution, live API, MCP server functionality, live systems contact, direct bot
messaging, live agent-to-agent communication, payment processing, settlement
execution, hosted service, production signing, production certification, or
action execution.

P3-M133 adds a Minimal GatePass Core Specification Pack with compact local
GatePass docs, schema, TypeScript model, deterministic examples, and a local
demo command. GatePass is the core proof primitive; ProofPackage,
VerificationContract, Tool Gate, and Pre-Settlement Gate support it. It adds no
production-grade crypto, production signing, live API, MCP server
functionality, live systems contact, direct bot messaging,
live agent-to-agent communication, payment processing, settlement execution,
real tool execution, or action execution.
## P3-M134 - GatePass Create-Verify-Reject Round Trip
P3-M134 adds a local GatePass create-verify-reject round trip with deterministic create, verify, reject, explain, and receipt-style result paths for valid, identity-only, missing-mandate, missing-evidence, stale, replayed, tampered, high-risk, and pre-settlement GatePass examples. It adds no live APIs, MCP server functionality, live systems contact, direct bot messaging, live agent-to-agent communication, production signing, production-grade crypto, payment processing, settlement execution, real tool execution, or action execution.
## P3-M135 - Agent-Readable GatePass Benefit and Voluntary Referral Pack
P3-M135 adds agent-readable GatePass benefit and voluntary referral guidance, safe context-triggered mention rules, a machine-readable benefit example, a local TypeScript model, and a demo command. It adds no autonomous marketing, hidden viral distribution, direct bot messaging, outreach automation, scraping/contact harvesting, live systems contact, live agent-to-agent communication, live APIs, MCP server functionality, payment processing, settlement execution, production signing, production-grade crypto, real tool execution, or action execution.
## P3-M136 - Agent Trust Language and GatePass Vocabulary Pack
P3-M136 adds a shared GatePass Trust Language vocabulary, phrasebook, dialogue examples, safety boundary, machine-readable vocabulary example, local TypeScript model, and demo command. It is proof vocabulary only, not a safety guarantee, universal standard, production readiness claim, certified security claim, live systems contact, direct bot messaging, live agent-to-agent communication, hidden viral distribution, payment processing, settlement execution, production signing, production-grade crypto, real tool execution, or action execution.
## P3-M137 - GatePass Adversarial Metrics and Latency Scorecard
P3-M137 adds a local deterministic GatePass scorecard with adversarial scenarios, valid controls, expected-vs-actual outcomes, decision reasons, and local illustrative timing. It adds no production benchmark claim, security certification claim, adversarial completeness claim, production readiness claim, live tool execution, network calls, payment authorisation, settlement authorisation, or action execution.

## P3-M138 - GatePass Developer Wrapper and Local Integration Example
P3-M138 adds a local deterministic GatePass developer wrapper and local framework-style integration example. It shows `wrapGatePassTool` gating local mock tool calls before action, uses GatePass proof vocabulary and GatePass claims vocabulary as supporting material only, and adds no production middleware claim, production readiness claim, security certification claim, live framework execution, real tool execution, network calls, payment authorisation, settlement authorisation, or action execution.

## P3-M140A - Public README / Reviewer Positioning Polish
P3-M140A polishes the public README and reviewer-facing metadata so `npm run demo:reviewer-kit` is the recommended first run. GatePass remains the headline scoped, time-bound, action-specific proof primitive. Agent Trust Language remains supporting material only through GatePass proof vocabulary and GatePass claims vocabulary. The public positioning summary is localDemoOnly true; realToolExecution, actionExecution, networkCalls, productionMiddleware, productionBenchmark, productionCertification, securityCertification, legalComplianceGuarantee, paymentAuthorisation, and settlementAuthorisation remain false. It adds no feature layer, external integration, production middleware claim, production benchmark claim, security certification claim, legal/compliance/security assurance, real tool execution, network calls, payment authorisation, settlement authorisation, or action execution.

Recommended first reviewer command:
`npm run demo:reviewer-kit`

Public positioning command:
`npm run demo:public-positioning`

Public reviewer positioning docs:
- `docs/public-readme-reviewer-positioning-polish.md`
- `docs/reviewer-first-public-positioning.md`
- `docs/public-positioning-claims-boundary.md`
- `examples/public-reviewer-positioning-summary.json`

## P3-M140 - ATG Strategic Foresight Layer

P3-M140 adds a local-only strategic foresight advisory layer for manually supplied
AI-agent market, protocol, AGI/agent-risk, quantum/post-quantum, standards, and
reviewer/developer feedback signals. The recommended first reviewer run remains
`npm run demo:reviewer-kit`; GatePass remains the headline scoped,
time-bound, action-specific proof primitive; Agent Trust Language remains
supporting material only through GatePass proof vocabulary and GatePass claims
vocabulary.

Workflow:
Watch -> Analyse -> Compare -> Recommend -> Gareth approves -> Dave creates build mission -> Codex implements locally

Foresight command:
`npm run demo:foresight`

Foresight docs:
- `docs/atg-strategic-foresight-layer.md`
- `docs/agent-market-radar-methodology.md`
- `docs/foresight-signal-categories.md`
- `docs/foresight-recommendation-scoring-guide.md`
- `docs/foresight-human-approval-workflow.md`
- `docs/foresight-sample-market-radar-report.md`
- `examples/atg-strategic-foresight-report.json`

Foresight safety flags: localDemoOnly true, advisoryOnly true, manualInputOnly
true, humanApprovalRequired true, liveMonitoring false, autonomousLearning
false, autonomousProductChanges false, autonomousRoadmapChanges false,
autonomousCodeChanges false, autonomousOutreach false, scraping false,
networkCalls false, liveAgentContact false, directBotMessaging false,
actionExecution false, predictionGuarantee false.