Workflow Commands
Workflow commands drive the core phase pipeline in PP Core: setting up a project, talking through a phase, planning it, building it, validating it, and shipping it. Each command operates on a numbered phase and moves the project one step further down the spec-driven loop.
/pp-new-project
Bootstraps a brand-new project by collecting deep context and producing the initial PROJECT.md, requirements, and roadmap.
/pp-new-project
/pp-new-project --auto @brief.md
| Flag | Argument | Behavior |
|---|---|---|
--auto | @file | Runs non-interactively, reading the supplied brief and chaining research, requirements, and roadmap generation without prompting. |
/pp-discuss-phase
Surfaces the open questions ("gray areas") for a phase and works through them with adaptive questioning before any planning begins.
/pp-discuss-phase 2
/pp-discuss-phase 3 --auto
/pp-discuss-phase 4 --power
/pp-discuss-phase 5 --assumptions
| Flag / Argument | Behavior |
|---|---|
N | The phase number to discuss (required). |
--all | Resolves every identified gray area within a single session. |
--auto | Answers gray areas automatically from context already on file. |
--batch | Clusters related gray areas so they can be addressed together. |
--analyze | Performs a deeper codebase scan before presenting the gray areas. |
--power | Switches to an extended questioning protocol intended for complex phases. |
--assumptions | Lists the implicit assumptions for the phase rather than starting a discussion. |
/pp-ui-phase
Produces a UI design contract for phases that include frontend work, capturing the visual and interaction specification ahead of planning.
/pp-ui-phase 3
| Argument | Behavior |
|---|---|
N | The phase number that needs a UI specification. |
/pp-plan-phase
The central planning step: researches the phase, drafts an executable PLAN.md, and verifies it before execution.
/pp-plan-phase 1
/pp-plan-phase 2 --skip-research
/pp-plan-phase 3 --research
/pp-plan-phase 4 --mvp --tdd
/pp-plan-phase --research-phase 2
| Flag / Argument | Behavior |
|---|---|
N | Phase number to plan; if omitted, the next unplanned phase is detected automatically. |
--research | Forces a fresh research pass even if cached research exists. |
--skip-research | Bypasses the research step entirely. |
--research-phase N | Runs research only for phase N, producing no plan. |
--gaps | Re-plans solely to close verified gaps from a prior run. |
--prd filepath | Reads a PRD file into planning context. |
--ingest glob | Pulls in ADR files matching the glob as additional context. |
--mvp | Plans the phase as a vertical MVP slice broken into feature increments. |
--tdd | Organizes the plan test-first. |
--reviews | Folds in feedback gathered from cross-AI review. |
--skip-verify | Skips the plan-checker verification step. |
/pp-plan-review-convergence
Runs repeated cross-AI review cycles, re-planning until high-severity concerns are cleared.
/pp-plan-review-convergence 3
/pp-plan-review-convergence 3 --gemini --codex
/pp-plan-review-convergence 3 --all --max-cycles 5
| Flag / Argument | Behavior |
|---|---|
N | Phase number to converge. |
--codex | Uses the Codex CLI as the reviewer (the default). |
--gemini | Uses the Gemini CLI as a reviewer. |
--claude | Spins up a separate Claude CLI session as a reviewer. |
--opencode | Uses OpenCode as a reviewer. |
--ollama | Uses a local Ollama server. |
--lm-studio | Uses a local LM Studio server. |
--llama-cpp | Uses a local llama.cpp server. |
--all | Engages every available CLI and local model. |
--max-cycles N | Caps the number of replan cycles (default: 3). |
/pp-execute-phase
Executes every plan in a phase, parallelizing independent work into waves.
/pp-execute-phase 2
/pp-execute-phase 3 --wave 1
/pp-execute-phase 4 --gaps-only
/pp-execute-phase 5 --interactive
| Flag / Argument | Behavior |
|---|---|
N | Phase number to execute. |
--wave N | Runs only the specified wave of the phase. |
--gaps-only | Executes only the gap-closure plans. |
--interactive | Runs plans sequentially, pausing at user checkpoints. |
/pp-verify-work
Walks through built features in a conversational UAT, auto-diagnosing failures it encounters.
/pp-verify-work 3
/pp-verify-work
| Argument | Behavior |
|---|---|
N | Phase number to verify; if omitted, you are prompted to choose one. |
/pp-ship
Opens a pull request for a verified phase and stages it for merge.
/pp-ship 4
/pp-ship 4 --draft
| Flag / Argument | Behavior |
|---|---|
N | Phase number to ship. |
--draft | Creates the PR as a draft rather than ready for review. |
/pp-progress
The unified status-and-routing command: reports where the project stands and points to (or runs) the next sensible action.
/pp-progress
/pp-progress --next
/pp-progress --next --auto
/pp-progress --do "add rate limiting to the API"
/pp-progress --forensic
| Flag | Behavior |
|---|---|
--next | Advances automatically to the next logical step. |
--next --auto | Chains steps without stopping until a blocking decision is reached. |
--do "description" | Maps a plain-language request to the most appropriate PP Core command. |
--forensic | Adds a six-check integrity audit to the status report. |
/pp-autonomous
Runs the full remaining pipeline across phases without manual intervention, discussing, planning, and executing each in turn.
/pp-autonomous
/pp-autonomous --from 3
/pp-autonomous --from 2 --to 5
/pp-autonomous --only 4
/pp-autonomous --interactive
| Flag | Behavior |
|---|---|
--from N | Begins the run at phase N. |
--to N | Stops the run after phase N. |
--only N | Processes only phase N. |
--interactive | Runs the discuss step inline while dispatching plan and execute work in the background. |
--text | Uses plain-text lists in place of interactive menus. |