README.md

Jentic One documentation

What Jentic One is and how it works: repository README.

Where do I go?

• Trying it out → README quickstart.
• Installing and running it (operator) → Install, then Secure for the security posture.
• Keeping it healthy (upgrades, backups, monitoring) → Operate.
• Using it (connecting agents, brokering calls) → Use.
• Understanding how it's built (surfaces, processes, data model) → Understand.
• Developing on it (spinning it up from source, extending the code) → Develop.
• You are an AI agent executing a task → agent runbooks and llms.txt.

Install

• Installation overview — artifacts, verification, air-gapped transfer
• Platform support — what runs on Linux / macOS / Windows / WSL2
• Windows — the WSL2 + native-CLI path, step by step
• CLI binaries — jentic / jenticctl download matrix and verification
• Docker
• docker-compose
• systemd
• Helm
• AWS Marketplace
• Agent runbook: install — the same install, written for an AI agent to execute

Operate

• Operating Jentic One — the day-2 map
• Monitoring & logs — health endpoints, log sinks, executions and the audit trail, metrics
• Upgrades — the version/migration contract, per-install pointers
• Backup & restore — what a restorable backup must contain
• Troubleshooting — symptom index for humans, backed by the agent failure catalogue
• Agent runbook: operate — status/logs/upgrade/uninstall as executable steps

Use

Get going:

• First brokered call — the six-step route map

Connect agents:

• Connect an agent — the four integration paths
• Run coding agents in isolation — the jentic run sandbox
• Serve MCP over HTTP — the optional hosted /mcp endpoint

Operate the catalog:

• Fix a spec with an overlay — patching imported API descriptions
• How credential resolution works — how secrets stay server-side, encrypted, and audited

Integrate apps:

• Register an OAuth client — third-party apps authenticating users
• CLI README — the full jentic / jenticctl command surface

Agent runbook: use — discover → request access → execute, as executable steps.

Secure

• Deploying Jentic One securely — threat model and the deployment-tier ladder
• Same-host setups — the threat model when an agent shares a machine with the instance, and the menu of isolation options (including jentic run)
• Agent runbook: harden — posture checks as executable steps

Understand

• Architecture overview — the five surfaces, two processes, three databases, one diagram
• Surfaces and layering — how the code is organized and the import rules that keep it that way
• Composition and processes — boot sequence, JENTIC__APPS, deployment topologies, background work
• Broker execution — the life of a brokered call: pipeline, resilience, egress, credential injection
• Data model — three databases, no cross-database foreign keys, immutable revisions
• Identity and authorization — actors, tokens, scopes, and the default-deny chain

Develop

• development/README.md — the folder index
• Local development setup
• Extending Jentic One
• Context and configuration
• Releasing
• Product scope — the product-fit rubric the issue-intake harness scores against
• deploy/README.md — build architecture: images, charts, Terraform, multi-arch
• tools/README.md — the CLI source installer and the generators behind the generated references
• Helm charts & local cluster — chart docs, kind workflow, smoke tests, observability stack
• AWS Marketplace publishing — the seller/maintainer side of the listing

Reference (generated — never hand-edit)

• Configuration — every config key, default, and env var (make config-reference)
• Endpoints & scopes — every HTTP route and its required scope (make endpoints)
• reference/README.md — how the generated material works