- Home
- Skills
- Agents & Orchestration
- ZCode Executor Pro
Works with the AI Agents you already use
ZCode Executor Pro
- **code-reviewer**: runs in an independent review session, checks strictly against the task contract and the diff, and grades findings as blocker / suggestion / nit.
$9
ZCode Executor Pro
Example session with this skill installed
I need to add a new REST endpoint to our Node.js auth service. Recommend a routing path and role preset for this.
- Read your context and instructions
- Compiled the zcode executor pro
I have analyzed your intent to add a REST endpoint to the auth service. Based on the deterministic routing table, I recommend the following configuration for your task. ## Routing Recommendation | Component | Value | | :--- | :--- | | **Recommended Presets** | `backend`, `minimal-change` | | **Target Template** | `templates/task-node-service.md` | | **Rationale** | Intent involves service-layer logic; backend role provides API discipline. | | **Route Source** | `table` | | **Read Only** | `false` | **Next steps** 1. Initialize the task file: `cp templates/task-node-service.md ~/.zcode-executor/worktrees/main/tasks/T-auth-endpoint.md` 2. Inject the `backend` role preset into the Background section of the task file. 3. Dispatch to the executor: `zcode-executor send --task tasks/T-auth-endpoint.md`
Connects securely to your tools. The creator never sees your data.
About this skill
### Closed-loop workflow Plan → run in a local executor session → evidence gate (git diff + tests) → review in an independent session → rework in the same executor session. Acceptance is based on evidence, not self-reports. ### 9 presets Compact role instructions injected into task files, not a full "agency" roster: - **general** (default): starts work even when the request is loose. It restates the goal in one sentence, records its assumptions, and doesn't keep stopping to ask questions. Switch to a specialised preset any time. - **minimal-change**: the smallest diff that meets acceptance, with no drive-by refactors and no scope creep. - **code-reviewer**: runs in an independent review session, checks strictly against the task contract and the diff, and grades findings as blocker / suggestion / nit. - **frontend**: touches only the listed routes and components, follows your existing design system, and covers accessibility basics such as keyboard paths, labels and contrast. - **backend**: changes only the listed endpoints and modules, and keeps your existing auth and input-validation patterns. - **rapid-prototype**: built for spikes, hackathons and throwaway proofs of concept. Getting a working demo comes first. - **architect**: design first. It writes a short ADR-style decision, then implements to that decision. - **evidence-qa**: accepts only real command output and git diff, and records the commands run, their exit codes, and whether acceptance was met. - **orchestrator**: a planner-side checklist. One independently verifiable change per task file, with the execute → review → rework rhythm kept on track. ### 6 task templates - **task-frontend**: frontend changes - **task-node-service**: Node services (endpoints, workers, CLIs) - **task-dsh-plugin**: plugin adaptation when the host platform or plugin API changes - **task-review**: independent review - **task-rework**: rework in the same executor session - **task-general**: general-purpose default ### Automatic routing Picks a preset and template from your request. A deterministic lookup table makes the call, and with a Jev key, Jev can assist. Requests recognised as read-only are handled read-only, with no file changes. ### Guided Jev first-run setup On first use, your host agent walks you through Jev pre-screen setup step by step: enter your key → a bundled script merges it into your config safely (never overwrites existing config, sets file permissions to 0600, refuses symlinks) → run doctor to check. You can skip it at any time. ### Still free and open source The core engine, CLI and base skill (Apache-2.0); Coding Plan's ~150% equivalent quota (not Pro-exclusive); the Jev pre-screen capability itself. ### After install Run `zcode-executor doctor` to confirm your setup, open this skill in your host agent, configure Jev if you want it, then take one small task through a full execute + review round.
Requires global zcode-executor npm package. Compatible with 0.1.x engine versions. Jev assist requires an external API key. Limited to nine role presets and six task templates.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 27 days ago
Requires