Devory VS Code Extension

The Devory extension turns VS Code into a practical control surface for Devory.

It helps you initialize a workspace, manage tasks, inspect doctrine and skills, run Devory, and review outputs without leaving the editor.

Why It Matters

Devory is not just a chat window for code generation. The extension exposes the actual workflow:

task lifecycle
doctrine and skill files
first-run workspace bootstrap
run execution and artifact inspection
review actions for supervised delivery

First Run

Open a repository, install the extension, and let Devory check whether the workspace is already initialized.

If it is not, the extension can scaffold it for you. That bootstrap flow works even without a global devory install.

Main Commands

Command	Description
`Devory: Initialize Workspace`	Create the Devory workspace structure
`Devory: List Tasks`	Open a task picker grouped by lifecycle stage
`Devory: Create Task`	Add a new task
`Devory: Generate Tasks from Idea`	Open the Task Builder webview, turn one idea into previewable task drafts, then commit selected drafts to backlog
`Devory: Move Task`	Move a task directly to another stage
`Devory: Promote Task`	Move work forward through the lifecycle
`Devory: Open Review Queue`	Jump into review-ready work
`Devory: Approve Review Task`	Approve a review task
`Devory: Send Review Task Back`	Return review work for rework
`Devory: Block Review Task`	Move a review task to blocked
`Devory: Requeue Blocked Task`	Move a blocked task back into the queue
`Devory: Archive Task`	Archive a task outside the normal active lanes
`Devory: Enrich Task`	Add missing structured sections to the active task file
`Devory: Add Acceptance Criteria`	Insert that section only if absent
`Devory: Add Verification Steps`	Insert that section only if absent
`Devory: Add Dependencies`	Insert that section only if absent
`Devory: Add Files Likely Affected`	Insert that section only if absent
`Devory: Start Factory Run`	Execute Devory from the bundled runtime
`Devory: Resume Factory Run`	Resume a failed or paused run
`Devory: Pause or Resume Factory Run`	Request a pause, or resume from the same control point
`Devory: Stop Factory Run`	Request an orderly stop for the active run
`Devory: Show Work`	Open live execution visibility (run state, heartbeat, doing/review focus)
`Devory: Inspect Recent Runs`	Open recent run records from the editor
`Devory: Show Routing Outcome Summary`	Summarize recent routing outcome ledger records in the Devory output channel
`Devory: Inspect Artifacts`	Browse saved execution outputs
`Devory: Governance Doctor`	Check workspace and CLI/runtime readiness
`Devory: Connect Cloud Account`	Start the local cloud-account connection flow
`Devory: Create Doctrine File`	Create a new doctrine file
`Devory: Create Skill`	Scaffold a reusable skill
`Devory: Create Agent`	Scaffold a new agent definition
`Devory: Archive Doctrine File`	Move doctrine into archive storage
`Devory: Archive Skill`	Move an old skill into archive storage
`Devory: Show Governance Status`	Report governance readiness and transport status
`Devory: Show Stored Data Locations`	Explain extension-owned local data versus project data
`Devory: Sweep the Workshop`	Delete only extension-owned local data
`Devory: Refresh Task Explorer`	Refresh task and governance views

Runtime Notes

Devory: Start Factory Run uses the packaged local runtime and shows an advisory dry-run/cost estimate summary before launch.
Run start now profiles ready tasks, applies routing policy, resolves concrete targets and adapters, and preserves selected versus actual execution metadata in the routing outcome ledger.
Devory: Resume Factory Run resumes an existing failed or paused run record.
Devory: Pause or Resume Factory Run and Devory: Stop Factory Run write local run-control signals that the active runtime checks between steps.
The extension exposes review and lifecycle commands directly, but those commands still operate against the repository state on disk.
Devory: Generate Tasks from Idea uses deterministic generation first, previews all generated drafts before any save, then offers post-commit handoff actions.
The Task Assistant can also run AI-assisted refinement against the active task, including local Ollama-backed refinement when that target is configured.
Devory: Show Work is a visibility/control surface; it does not replace governance command execution by devory worker.

Workspace Shape

The extension works directly against the repo:

tasks/ for work
runs/ for run records
artifacts/ for durable execution output
doctrine/ and skills/ for engineering guidance and reuse

Routing Outcome Review

Ledger path: artifacts/routing-outcomes/execution-outcomes.jsonl
Command: Devory: Show Routing Outcome Summary
Truth model: selected provider/target/adapter and actual provider/target/adapter stay separate
Cloud behavior: local-first by default; cloud-bound runs can require confirmation depending on routing policy

Devory.AI