// Execute all pending tasks in task.json using the zeus-runner.sh shell loop. Tasks are grouped into dependency waves and run with fresh claude CLI subagents. Each task gets an atomic git commit and a mandatory ai-log entry. Supports --version flag for v2/v3/v4 execution.
Brainstorm features or run full project planning through structured dialogue. Use --full for initial project-wide planning (includes north star metrics if not set), or --feature <name> for a single feature. Always asks one question at a time. Invokes zeus-plan automatically after spec is approved.
Map an existing (brownfield) codebase into deterministic Zeus artifacts under .zeus/{version}/ so downstream skills can reuse real project context.
Initialize a new Zeus project. Creates .zeus/main/ directory structure, sets up config.json with north star metrics, and writes the first evolution.md entry. Use this FIRST before any other zeus skill.
Convert a spec document into structured prd.json user stories and task.json execution units. Called automatically by zeus-brainstorm, but can also be invoked directly. Supports --version flag to target v2/v3/v4 folders.
Display a full status report of the Zeus project — all active versions, task progress, recent feedback attribution, evolution history, and next recommended action.
Create a new version folder (.zeus/v2/, .zeus/v3/ etc.) when user feedback reveals needs that can't be met by the main version. Inherits main config and declares differences. Records everything in evolution.md.
| name | zeus-execute |
| description | Execute all pending tasks in task.json using the zeus-runner.sh shell loop. Tasks are grouped into dependency waves and run with fresh claude CLI subagents. Each task gets an atomic git commit and a mandatory ai-log entry. Supports --version flag for v2/v3/v4 execution. |
Run pending tasks with dependency awareness, strict quality gates, and atomic commits.
.zeus/{version}/config.json exists..zeus/{version}/task.json exists and contains pending tasks unless filtered by --task..zeus/scripts/zeus-runner.sh exists and is executable.claude CLI is installed and reachable.--task T-###--wave <N>--version vNLoad pending tasks and group by wave.
Show preview summary before running:
Require user confirmation to proceed.
bash .zeus/scripts/zeus-runner.sh \
--task-file .zeus/{version}/task.json \
--log-dir .zeus/{version}/ai-logs \
--version {version}
Track runner output and report progress by wave:
After run, verify:
passes: true,commit_sha is filled,completed when all related tasks pass.Update relevant roadmap milestone status and write summary log.
Create .zeus/{version}/ai-logs/{ISO-ts}-execute.md.
Each task completion must include:
{type}({task_id}): {description}task.json update (passes, commit_sha, optional ai_log_ref).zeus/{version}/ai-logs/{ISO-ts}-{task_id}.md.zeus/{version}/progress.txtBrownfield extension:
refactor_of, switch to minimal-intrusion strategy:
refactor_of details in task-level AI log execution summary.>3 retries): stop current wave and request task split,/zeus:feedback or /zeus:status).After each wave completes, offer:
Wave N complete. Refresh test flows for completed tasks? [yes/no]
(Runs: bash .zeus/scripts/generate-tests.sh --version {version} --force)
This regenerates {platform}.test.json scenarios for tasks marked passes: true in the current wave, keeping test coverage aligned with shipped code.
Use zeus-executor to supervise multi-wave execution status and enforce completion gates.
Use zeus-tester (via /zeus:test-gen) after execution waves to generate or refresh platform test flows.