Agentic SDLC
Evidence-first engineering workflows for Visual Studio Code.
Agentic SDLC brings the runtime-backed parts of the broader Agentic SDLC Framework into VS Code: structured diagnostics, resumable session state, trace capture, and a workspace view over the execution runtime already living in your repo.
Why this extension exists
Most AI coding experiences are excellent at conversation, editing, and ad-hoc tool use. They are weaker at preserving execution evidence, resume context, and release-grade workflow state across sessions and across tools.
This extension focuses on the part that usually goes missing:
- Typed debug evidence for command, test, CI, and browser failures
- Workspace-scoped runtime state that survives editor restarts
- Trace markdown and event history stored in your repo, not trapped in chat history
- A thin VS Code shell over the same runtime modules used by the CLI and MCP layers
What you get
- Runtime view for active session, plan, trace, execution mode, approvals, task counts, and recent events
- Diagnostics view for failure summaries by kind and severity
- Capture commands for command failures, test failures, CI failures, and browser verification
- Resume snapshot command that opens the structured runtime state as JSON
- Latest trace command that jumps directly to the most recent markdown trace in
.ai/traces/
Architecture
The VS Code extension does not create a second workflow engine. It reuses the same workspace-scoped runtime used by the framework itself and activates on demand through its views and commands.
Control plane layers
- VS Code shell
Exposes views, commands, and status bar state inside the editor.
- Runtime bridge
Maps VS Code actions to the framework libraries.
- Workspace runtime
Reads and writes
.ai/session-state/, .ai/traces/, and diagnostics artifacts.
- Core framework
Shares logic with the CLI, MCP surface, and bundled runtime modules shipped with the extension package.
Where this fits
Compared with IDE-native agent features
VS Code 1.110 already offers strong native agent capabilities such as agent plugins, browser tools, session memory, /compact, rename/usages tools, and the Agent Debug panel (VS Code 1.110 release notes).
This extension is meant to complement that layer, not duplicate it.
What Agentic SDLC adds on top:
- Repo-persisted runtime state under
.ai/session-state/
- Typed evidence bundles that normalize command, test, CI, and browser failures
- Trace markdown and event history that remain available outside a single chat session
- A path toward release gates, review loops, and cross-repo impact workflows from the same control plane
Spec-first tools such as spec-kit are strong when the center of gravity is requirements, proposal quality, and implementation planning. Agentic SDLC is stronger when the center of gravity is execution evidence, workflow state, review readiness, and resumable delivery control.
Compared with chat-first extensions
Chat-first tools optimize for request/response speed. Agentic SDLC optimizes for:
- continuity
- auditability
- delivery state
- operational debugging
Feature map
| Capability |
IDE-native agent features |
Spec-first toolkits |
Agentic SDLC |
| Chat, editing, and built-in tools |
Strong |
Partial |
Uses VS Code surface |
| Persistent runtime state in repo |
Limited |
Partial |
Strong |
| Typed failure evidence |
Limited |
Weak |
Strong |
| Resume snapshots and trace files |
Limited |
Weak |
Strong |
| Release-gate evolution path |
Limited |
Partial |
Strong |
| Cross-tool portability (CLI + MCP + editor) |
Limited |
Partial |
Strong |
Quick start
- Open a repository that already contains Agentic SDLC runtime files, or scaffold the framework first from the main repo:
npx agentic-sdlc-development init .
- Install the extension from Marketplace or with:
code --install-extension PramodKumarVoola.agentic-sdlc-vscode
The packaged extension syncs and bundles the runtime modules it needs for Marketplace-style installs and .vsix distribution.
- Open the Command Palette and run one of:
Agentic SDLC: RefreshAgentic SDLC: Capture Command FailureAgentic SDLC: Capture Test FailureAgentic SDLC: Capture CI FailureAgentic SDLC: Capture Browser VerificationAgentic SDLC: Show Resume SnapshotAgentic SDLC: Open Latest Trace
Example workflow
- A test run fails locally.
- Run
Agentic SDLC: Capture Test Failure.
- The extension records structured diagnostics and a trace into your workspace.
- Open the Runtime and Diagnostics views to inspect the updated state.
- Use the trace and event files in review, CI triage, or follow-up automation.
Settings
| Setting |
Default |
Purpose |
agenticSdlc.eventsLimit |
10 |
Number of runtime events shown in the Runtime view |
agenticSdlc.autoRefresh |
true |
Refresh views after capture commands complete |
Requirements
- VS Code
1.110+
- A workspace using the Agentic SDLC runtime layout
- a workspace using the Agentic SDLC runtime layout
Roadmap direction
This release is intentionally debug-first. The next layers are designed to grow from the same control plane:
- richer review and release gating
- approval queue operations
- cross-repo impact views
- plugin-pack visibility
- deeper MCP-assisted workflows
Source
MIT licensed.
