/wbActOn — Live Demo ()
This is what /wbActOn actually does on wb-labs as the workspace stands today (2026-05-05). The matrix below mirrors the exhaustive simulation, but every cell is filled from the live state of the repo — real audit findings, real model recommendations, real memory items.
1. Live target
| Field | Live value |
|---|---|
| Most recent audit | packages/wb-core/.agents/workflows/reports/2026/05/04/audits/audit_wb-core_20260504.md |
| Audit findings | 23 items across 8 sections |
| Active triage model | the AI agent via Antigravity |
| Existing action files | None for today (would be Entry #1) |
| Parked memory items | project_pkg_dist_mismatch.md, project_wbdataviewer_apiResponse.md, project_docs_edition.md |
2. What each argument resolves to today
| Argument | Live resolution |
|---|---|
/wbActOn audit_wb-core_20260504.md | Mirror mode. Reads the audit, writes action to reports/2026/05/05/actions/action_wb-core_20260505.md. |
/wbActOn packages/wb-core | Pick mode. Lists recent reports → finds the 20260504 audit → asks to confirm. |
/wbActOn competitor_analysis.md | Side-car mode. Anchors to core2/.agents/workflows/reports/2026/05/05/actions/. |
/wbAudit packages/wb-core --act | Chained: runs a fresh audit then pipes it through /wbActOn. |
3. Per-flag behavior, applied live
| Flag | If invoked now |
|---|---|
/wbActOn audit_wb-core_20260504.md | Creates action file (Entry #1). §0 ranked thread + per-section annotations. |
/wbActOn audit_wb-core_20260504.md --wbPlan | Action file + companion plan with task tables for each 🔵 finding. |
/wbAudit packages/wb-core --act --wbPlan | Fresh audit → action → plan. Three files in one invocation. |
4. Pipelines on this exact workspace
/wbActOn Live Execution Simulator
Pipeline A — Triage yesterday's audit
/wbActOn audit_wb-core_20260504.md> /wbActOn audit_wb-core_20260504.md[SYSTEM] Ingesting audit_wb-core_20260504.md...
[TYPE] Audit (23 findings, 8 sections).
[CHECK] action_wb-core_20260505.md → does not exist. Creating Entry #1.
[WROTE] reports/2026/05/05/actions/action_wb-core_20260505.md
Generated Output (
| Rank | Finding | Color | Time | Model | Justification |
|---|---|---|---|---|---|
| 1️⃣ | tierEnforcement.js alg: "none" bypass | 🔴 | TODAY | the agent 4 | Security — any JWT with "none" alg passes the gate |
| 2️⃣ | renderString XSS vector | 🔴 | TODAY | the agent 4 | Input sanitization — user-facing risk |
| 3️⃣ | Dead handlers in WBC.js (3 unused) | 🟡 | THIS WEEK | Qwen3 Coder | Clean sweep, low risk |
| 4️⃣ | dist-folder mismatch in wbc-ui2-cdn | 🔵 | THIS WEEK | the agent 4 | Multi-step — package.json + CDN consumers |
⚠️ Note: Focus on ranks 1-2 today. ROI estimate: Today's 2 ranks move the audit security score from ~60 → ~85 with ~3 hours of effort.
5. What would refuse today
| Trigger | Live response |
|---|---|
/wbActOn with no argument | ❌ Provide a file or folder path. |
/wbActOn freeform_notes.txt (unstructured) | ❌ Cannot triage unstructured content. Source needs sections, tables, or numbered items. |
/wbActOn plan_wb-core_20260504.md --wbPlan | ❌ Source is already a plan. Use --act alone to re-rank. |
| Callout without pasteable command | Self-correction violation. Every annotation must include the exact /wb* command or one-shot prompt. |
/wbActOn on a pure context.md | ℹ️ No actions — this is a state snapshot. Use it as input to /wbAudit or /wbPlan, not as a task list. |
The pattern: /wbActOn converts diagnosis into execution order. It's the bridge between "what's wrong" (audit/review/standup) and "what to do about it" (/wbWork, /wbDebug, /wbPlan). It never invents, always annotates, and always tells you which model should do the work.