@hanzlaa/rcode 3.6.16 → 4.0.0

package/rcode/brain/README.md

@@ -0,0 +1,38 @@
+# rcode/brain — The rcode Context Layer
+
+This directory is populated on install (and on every `/rcode-update`) by `rcode-tools brain pull`. Its job is simple: pull rcode's institutional knowledge — PR/commit/issue standards, architecture decisions, internal guides — into every rcode user's project so their AI assistant already knows how rcode builds.
+
+## Structure
+
+```
+rcode/brain/
+├── sources.yaml           # upstream repo configuration (see below)
+├── rcode-github/          # PR/commit/issue standards from the rcode GitHub org
+├── rcode-docs/            # architecture, guides, playbooks from the rcode docs repo
+└── best-practices/        # in-repo best practices, pulled from rcode/skills/_shared/
+```
+
+## Sources
+
+Content is pulled from three kinds of source:
+
+1. **External rcode repos** — GitHub org + docs repo. URLs are placeholders until M5 / issue #162 lands with the real addresses. Once filled, `brain pull` uses `git` sparse-checkout to fetch only the paths listed in `sources.yaml`.
+2. **Self** — the special `repo: self` entry points back at this repo, so in-tree best-practices land under `rcode/brain/best-practices/` on every install.
+3. **Cached** — fetched content is cached locally with a short TTL (configurable per source, default 6h). `brain pull` respects the cache; pass `--fresh` to force a re-clone.
+
+## Update policy
+
+- Local edits to any file under `rcode/brain/` are **overwritten** on every `/rcode-update`. rcode treats the upstream sources as the single source of truth. If the rcode standard needs to change, the change happens upstream — then every rcode user benefits.
+- Per-invocation: `node .rcode/bin/rcode-tools.cjs brain pull`.
+- Per project: runs automatically as part of `npx rcode install`.
+
+## Privacy
+
+If any source is private, `brain pull` uses the user's `gh auth` token (set `defaults.private: true` in `sources.yaml`). If `gh auth status` fails, the pull skips that source with a clear message instead of hanging.
+
+## Status — v2.0
+
+- `sources.yaml` is scaffolded with placeholder URLs.
+- `brain pull` subcommand is wired in `rcode/bin/rcode-tools.cjs`.
+- Real URLs arrive with issue #162 (M5).
+- Live MCP replacement lands in v3.0 (issue #163 / design doc at `docs/adr/mcp-design.md`).

package/rcode/brain/best-practices/no-autonomous-bypass.md

@@ -0,0 +1,37 @@
+# No Autonomous-Mode Bypass
+
+This fragment is referenced by every step-file workflow that enforces halt-at-menu. Do not inline its contents into individual step files — reference this file so the rule stays single-sourced.
+
+## The Rule
+
+Step-file skills (`rcode-create-prd`, `rcode-sprint-planning`, `rcode-create-epics-and-stories`, etc.) halt at every menu and wait for user input. **This is invariant.** The agent MUST NOT invent an "autonomous mode", "research mode", or any other self-declared bypass that skips discovery questions.
+
+Phrases like *"just write it autonomously"*, *"create the full thing ready to execute"*, *"skip the questions"*, or *"use research mode"* from the user are **not** authorization to bypass halt. They are signals that the user wants the skill to move efficiently — answer by running the next step's questions compactly, not by skipping them.
+
+## The Only Sanctioned Bypass Paths
+
+There are exactly two ways to legitimately run a skill without halting at menus:
+
+1. **Project-wide:** `.rcode/config.yaml` → `mode: yolo`. The config loader flags `yoloMode=true` in the runtime context; step files check this flag and may auto-advance.
+2. **Per-invocation:** `/rcode-do --auto <question>` on the router. The router sets `autoMode=true` and passes it to the dispatched skill.
+
+If **neither** flag is set, halt is mandatory regardless of what the user's prompt text says.
+
+## Required Behavior When Asked to Bypass Without a Flag
+
+1. Acknowledge the user's intent in one sentence.
+2. Show the current step's menu anyway.
+3. Tell the user the two sanctioned paths (config flag or `--auto`) if they want fully hands-off execution.
+4. Do not generate any output artifact (PRD, sprint plan, roadmap, etc.) until the user either selects a menu option or sets one of the sanctioned flags.
+
+## Anti-Example (What NOT to Do)
+
+> User: *"use rcode research skills and create best PRD ready to execute"*
+>
+> ❌ Agent: *"Switching to autonomous mode — researching the domain and drafting a complete execution-ready PRD."* *(then writes full PRD)*
+
+That response invents a mode that does not exist and violates the halt invariant.
+
+## Correct Response
+
+> ✅ Agent: *"I hear you want the PRD written end-to-end. The halt rule applies unless you set `mode: yolo` in `.rcode/config.yaml` or re-invoke via `/rcode-do --auto`. Here is the step-01 menu — pick Continue and I will drive each step concisely."*

package/rcode/brain/best-practices/no-theoretical-suggestions.md

@@ -0,0 +1,56 @@
+# No Theoretical Suggestions Rule
+
+Referenced by all agents that propose plans, file changes, or implementation steps.
+
+## The Rule
+
+**Every suggestion, plan, or implementation step must be grounded in what actually exists in the codebase.** Never propose a change based on assumed structure, naming conventions, documentation, or what "probably" exists.
+
+## What This Means in Practice
+
+Before including any file path, function name, config key, or field in a plan:
+
+1. Confirm it exists — use `Read`, `grep`, `find`, or `glob`.
+2. If it doesn't exist, say so explicitly: "this doesn't exist yet."
+3. Only then decide whether to create it (if the user asked) or report the gap (if they didn't).
+
+## The Test
+
+Could every line of your plan be traced to a specific file and line number in the repo right now? If not, the plan contains theoretical content — do not ship it.
+
+## Failure Pattern to Avoid
+
+- Agent reads an investigation doc or workflow description.
+- Agent produces a plan based on what the doc *says* should exist.
+- Agent never verifies the described files/fields/functions actually exist.
+- Plan references symbols that are missing, misnamed, or not yet implemented.
+- Execution fails or produces wrong results.
+
+This pattern is forbidden. Investigation documents describe intent, not ground truth. Code is ground truth.
+
+## Correct Behavior
+
+```
+# WRONG — theoretical
+"Fix the force_web_search toggle in backend/config.py"
+(without checking if force_web_search actually exists in that file)
+
+# RIGHT — grounded
+grep -r "force_web_search" backend/
+→ no results
+→ "force_web_search doesn't exist in the codebase yet. The investigation doc was theoretical. Here's what actually exists: ..."
+```
+
+## When You Can't Find Something
+
+- Do NOT assume it's in a different file or under a different name.
+- Do NOT write a plan that assumes you'll find it later.
+- STOP and report: "I searched for X and found nothing. Either it doesn't exist yet or the investigation was theoretical."
+
+## Applies To
+
+- `rcode-planner` — every task in SPRINT.md must reference real files
+- `rcode-executor` — read the target file before writing any change
+- `rcode-phase-researcher` — cite actual code paths, not presumed ones
+- `rcode-sprint-checker` — flag plans that reference non-existent symbols
+- All council agents — proposals must name real constraints, not theoretical ones

package/rcode/brain/best-practices/research-citation-rule.md

@@ -0,0 +1,39 @@
+# Research & Citation Rule
+
+Referenced by any step-file workflow that writes content derived from external sources (competitor pricing, API tiers, market sizing, benchmark numbers, vendor claims, etc.).
+
+## The Rule
+
+If an output artifact (PRD, research note, architecture doc, roadmap) contains a specific external claim — a price, a percentage, a named competitor feature, an API limit — then **every such claim must be traceable to a `WebFetch` (or equivalent document-fetch) call made in the same session**. `WebSearch` snippets alone are not sufficient evidence because snippets are paraphrased and may be stale.
+
+## What This Means in Practice
+
+Before writing a claim like *"Competitor X costs $12.50/mo"*:
+
+1. Run `WebFetch` on the source URL (not just `WebSearch` for a snippet).
+2. Quote the exact sentence from the fetched content in your internal reasoning.
+3. Include the URL in the citations block of the artifact.
+4. If the URL is paywalled, login-walled, or returns a different claim than expected, drop the number — do not paraphrase from memory or search-result previews.
+
+## The Trap to Avoid
+
+Observed failure pattern (rcode session, social-poster-x PRD generation):
+
+- Agent ran two `WebSearch` calls.
+- Agent wrote a PRD citing four specific URLs with specific dollar amounts.
+- Agent never ran `WebFetch` on any of the four URLs.
+- Numbers may be accurate-looking hallucinations.
+
+This pattern is forbidden. A citation block without `WebFetch` evidence is a lie to the reader.
+
+## If the User Is Running Autonomously
+
+`mode: yolo` and `/rcode-do --auto` bypass halt-at-menu, but they do **not** bypass this rule. A fully-autonomous run still has to fetch every cited URL before writing the claim.
+
+## Checklist Before Writing a Citation
+
+- [ ] `WebFetch` was called on this URL in this session.
+- [ ] The fetched content contains the specific number / claim being cited.
+- [ ] The claim in the artifact matches the fetched content verbatim (or is a clearly labeled summary).
+
+If any box is unchecked: remove the claim, replace with *"(needs source)"*, and flag to the user.

package/rcode/brain/best-practices/state-sync-rule.md

@@ -0,0 +1,43 @@
+# State Sync Rule
+
+Referenced by skills that write `ROADMAP.md`, `epics.md`, or sprint artifacts. These artifacts are not authoritative on their own — downstream workflows (`/rcode-status`, `/rcode-progress`, `/rcode-execute`) read `.rcode/state.json`. If you write a planning artifact and skip state sync, the project ends up with two divergent pictures.
+
+## The Rule
+
+Immediately after appending content to any of:
+
+- `.planning/ROADMAP.md` — milestones and phases
+- `.planning/epics.md` — epics and stories
+- `.rcode/phases/{phase}/sprint-{N}.md` — sprint commitments
+
+Call the state-sync helper:
+
+```bash
+node .rcode/bin/rcode-tools.cjs state sync --from-disk
+```
+
+or, for a more targeted write, use the JSON-API helpers exposed by `rcode-tools.cjs`:
+
+```bash
+node .rcode/bin/rcode-tools.cjs state upsert-milestone <json>
+node .rcode/bin/rcode-tools.cjs state upsert-phase <json>
+node .rcode/bin/rcode-tools.cjs state upsert-epic <json>
+```
+
+If `rcode-tools.cjs` does not yet expose the needed subcommand, fall back to `state sync --from-disk` which re-parses `ROADMAP.md` + `epics.md` and rebuilds the relevant sections of `state.json`.
+
+## Verification After Sync
+
+- `node .rcode/bin/rcode-tools.cjs state read` returns a phase count that matches the phase table in `ROADMAP.md`.
+- `/rcode-status` and `/rcode-progress`, run back-to-back, agree on the current milestone name and phase count.
+
+## Why This Matters
+
+Observed failure (rcode, social-poster-x install, Apr 2026):
+
+- User ran `/rcode-create-epics-and-stories`.
+- `ROADMAP.md` gained 10 phases; `epics.md` gained 62 stories.
+- `.rcode/state.json` remained at the initial bootstrap with 1 phase.
+- `/rcode-status` showed 1 phase; `/rcode-progress` showed 10.
+
+That divergence is what this rule exists to prevent.

package/rcode/brain/sources.yaml

@@ -0,0 +1,62 @@
+# rcode Brain — content sources
+#
+# This file lists upstream repositories whose content is pulled into
+# rcode/brain/ on install and on every `/rcode-update`. Anything pulled
+# here ends up in every rcode user's project context.
+#
+# Repo URLs are placeholders until M5 (issue #162) when Hanzla supplies
+# the real rcode GitHub org and docs repo addresses. `brain pull` will
+# detect `<PLACEHOLDER>` entries and skip them with a clear message.
+#
+# Real URLs land as part of issue #162. Until then, this file documents
+# the intent and shape of the pipeline.
+
+version: 1
+
+defaults:
+  # Branch to track when no per-source override is set.
+  branch: main
+  # How long before a cached checkout is considered stale and re-cloned.
+  cache_ttl: 6h
+  # Private repos — if true, `brain pull` uses the user's `gh auth` token.
+  # Set to false for public repos (faster, no auth needed).
+  private: true
+
+sources:
+  - name: rcode-github-standards
+    description: >
+      PR, issue, and commit standards for the rcode GitHub org. Every
+      rcode user's AI sees these when reviewing or drafting a PR.
+    repo: "<PLACEHOLDER: github.com/rcode-om/???>"
+    branch: main
+    paths:
+      - docs/pr-standards.md
+      - docs/issue-standards.md
+      - docs/commit-standards.md
+      - docs/review-checklist.md
+    dest: rcode-github/
+
+  - name: rcode-docs
+    description: >
+      Architecture decisions, internal guides, role playbooks. The bulk
+      of the rcode institutional knowledge lives here.
+    repo: "<PLACEHOLDER: github.com/rcode-om/???>"
+    branch: main
+    paths:
+      - "architecture/**/*.md"
+      - "guides/**/*.md"
+      - "standards/**/*.md"
+      - "playbooks/**/*.md"
+    dest: rcode-docs/
+
+  - name: rcode-best-practices
+    description: >
+      Hanzla's one-year-of-usage best practices — in-repo, version-
+      controlled alongside the rest of rcode. Resolved from the
+      installed .rcode/skills/_shared/ dir (or the package's own
+      rcode/skills/_shared/ when running inside the rcode repo).
+    repo: self
+    paths:
+      - .rcode/skills/_shared/**/*.md
+      - rcode/skills/_shared/**/*.md
+    dest: best-practices/

package/rcode/commands/add-phase.md

@@ -0,0 +1,18 @@
+---
+name: rcode-add-phase
+description: "Add a new integer phase to the end of the current milestone. Auto-calculates the next phase number, creates the phase directory, and updates ROADMAP.md. Use when scope expands mid-milestone."
+argument-hint: "<phase-name>"
+allowed-tools: Read, Write, Bash, Glob, Grep, AskUserQuestion
+---
+
+<objective>
+Execute add-phase workflow
+</objective>
+
+<execution_context>
+@.rcode/workflows/add-phase.md
+</execution_context>
+
+<process>
+Execute the add-phase workflow from @.rcode/workflows/add-phase.md end-to-end.
+</process>

package/rcode/commands/add-tests.md

@@ -0,0 +1,18 @@
+---
+name: rcode-add-tests
+description: Generate unit and E2E tests for a completed phase based on its SUMMARY.md, CONTEXT.md, and implementation. Classifies ea
+argument-hint: ""
+allowed-tools: Read, Write, Bash, Glob, Grep, AskUserQuestion, Agent
+---
+
+<objective>
+Generate unit and E2E tests for a completed phase based on its SUMMARY.md, CONTEXT.md, and implementation. Classifies ea
+</objective>
+
+<execution_context>
+@.rcode/workflows/add-tests.md
+</execution_context>
+
+<process>
+Execute the add-tests workflow from @.rcode/workflows/add-tests.md end-to-end.
+</process>

package/rcode/commands/add-todo.md

@@ -0,0 +1,8 @@
+---
+name: rcode-add-todo
+description: Capture an idea or task for later work
+argument-hint: "<todo-title>"
+allowed-tools: Bash, Read, Write
+---
+
+@.rcode/workflows/add-todo.md

package/rcode/commands/analyze-dependencies.md

@@ -0,0 +1,11 @@
+---
+name: rcode-analyze-dependencies
+description: Analyze phase dependencies, suggest "Depends on" entries for ROADMAP.md
+argument-hint: ""
+allowed-tools:
+  - Read
+  - Grep
+  - Bash
+---
+
+@.rcode/workflows/analyze-dependencies.md

package/rcode/commands/audit-fix.md

@@ -0,0 +1,14 @@
+---
+name: rcode-audit-fix
+description: Autonomous audit-to-fix pipeline — find issues, classify, fix, test, commit.
+argument-hint: "[--max N] [--severity high|medium|all] [--dry-run] [--source <audit>]"
+allowed-tools:
+  - Read
+  - Grep
+  - Glob
+  - Bash
+  - Task
+  - AskUserQuestion
+---
+
+@.rcode/workflows/audit-fix.md

package/rcode/commands/audit-milestone.md

@@ -0,0 +1,12 @@
+---
+name: rcode-audit-milestone
+description: Cross-phase audit — verify milestone completion against original goals
+argument-hint: "[--strict] [--report]"
+allowed-tools:
+  - Read
+  - Grep
+  - Bash
+  - AskUserQuestion
+---
+
+@.rcode/workflows/audit-milestone.md

package/rcode/commands/audit-uat.md

@@ -0,0 +1,18 @@
+---
+name: rcode-audit-uat
+description: Cross-phase audit of all UAT and verification files. Finds every outstanding item (pending, skipped, blocked, human_need
+argument-hint: ""
+allowed-tools: Read, Write, Bash, Glob, Grep, AskUserQuestion, Agent
+---
+
+<objective>
+Cross-phase audit of all UAT and verification files. Finds every outstanding item (pending, skipped, blocked, human_need
+</objective>
+
+<execution_context>
+@.rcode/workflows/audit-uat.md
+</execution_context>
+
+<process>
+Execute the audit-uat workflow from @.rcode/workflows/audit-uat.md end-to-end.
+</process>

package/rcode/commands/audit.md

@@ -0,0 +1,8 @@
+---
+name: rcode-audit
+description: Single audit entry point — asks what to audit (plans, phase, milestone, UAT, code, fix, work, lens) and dispatches to the right subroute. Honours .rcode/config.yaml mode.
+argument-hint: "[plans | phase | milestone | uat | code | fix | work | lens [<1-15> | all]] [--report] [...subroute args]"
+allowed-tools: Read, Write, Bash, AskUserQuestion
+---
+
+@.rcode/workflows/audit.md

package/rcode/commands/autonomous.md

@@ -0,0 +1,19 @@
+---
+name: rcode-autonomous
+description: Execute remaining phases autonomously with minimal human intervention. Runs plan → execute → verify cycles for unfinished work, pausing at checkpoints and failures.
+argument-hint: "[--from N] [--to M] [--only N] [--interactive]"
+allowed-tools: Read, Bash, Agent, AskUserQuestion
+---
+
+<objective>
+Execute remaining incomplete phases autonomously — plan, execute, verify in a loop, pausing only at checkpoints, failures, or decision gates.
+</objective>
+
+<execution_context>
+@.rcode/workflows/autonomous.md
+</execution_context>
+
+<process>
+Execute the autonomous workflow from @.rcode/workflows/autonomous.md end-to-end.
+Loop through todo phases: spawn rcode-planner if no SPRINT.md, then rcode-executor, handle checkpoints.
+</process>

package/rcode/commands/brainstorm.md

@@ -0,0 +1,11 @@
+---
+name: rcode-brainstorm
+description: Guided brainstorming session — select a method, apply it to your challenge, generate ideas systematically
+argument-hint: "<challenge> [--method=METHOD] [--people=N] [--personas=LIST]"
+allowed-tools:
+  - Read
+  - Bash
+  - AskUserQuestion
+---
+
+@.rcode/workflows/brainstorm.md

package/rcode/commands/capture.md

@@ -0,0 +1,12 @@
+---
+name: rcode-capture
+description: Capture a thought as todo (default), note, seed, or list pending — unified entry replacing /rcode-add-todo, /rcode-note, /rcode-plant-seed, /rcode-check-todos.
+argument-hint: "[text] [--note] [--seed] [--list]"
+allowed-tools:
+  - Read
+  - Write
+  - Bash
+  - Glob
+---
+
+@.rcode/workflows/capture.md

package/rcode/commands/chain.md

@@ -0,0 +1,8 @@
+---
+name: rcode-chain
+description: Run a sequential agent pipeline (research → scope → build). Each stage reads the previous stage's artifact.
+argument-hint: "<preset|agent-list> <topic>"
+allowed-tools: Read, Write, Bash, Agent
+---
+
+@.rcode/workflows/chain.md

package/rcode/commands/check-implementation-readiness.md

@@ -0,0 +1,18 @@
+---
+name: rcode-check-implementation-readiness
+description: "Verify a feature is fully ready to implement — PRD approved, architecture approved, dependencies identified — before writing code."
+argument-hint: "[--phase <n>]"
+allowed-tools: Read, Write, Bash, Glob, Grep, Agent
+---
+
+<objective>
+Execute check-implementation-readiness workflow
+</objective>
+
+<execution_context>
+@.rcode/workflows/check-implementation-readiness.md
+</execution_context>
+
+<process>
+Execute the check-implementation-readiness workflow from @.rcode/workflows/check-implementation-readiness.md end-to-end.
+</process>

package/rcode/commands/check-todos.md

@@ -0,0 +1,18 @@
+---
+name: rcode-check-todos
+description: List all pending todos, allow selection, load full context for the selected todo, and route to appropriate action.
+argument-hint: ""
+allowed-tools: Read, Write, Bash, Glob, Grep, AskUserQuestion, Agent
+---
+
+<objective>
+List all pending todos, allow selection, load full context for the selected todo, and route to appropriate action.
+</objective>
+
+<execution_context>
+@.rcode/workflows/check-todos.md
+</execution_context>
+
+<process>
+Execute the check-todos workflow from @.rcode/workflows/check-todos.md end-to-end.
+</process>

package/rcode/commands/checkpoint-preview.md

@@ -0,0 +1,13 @@
+---
+name: rcode-checkpoint-preview
+description: Human-in-the-loop change review — makes sense of a diff, focuses attention where it matters, and walks through testing. Use when you say "checkpoint", "walk me through this", or "human review".
+argument-hint: "[<branch-or-diff>]"
+allowed-tools:
+  - Read
+  - Bash
+  - Grep
+  - Glob
+  - AskUserQuestion
+---
+
+@.rcode/workflows/checkpoint-preview.md

package/rcode/commands/cleanup.md

@@ -0,0 +1,18 @@
+---
+name: rcode-cleanup
+description: "Archive completed milestone phase directories into .planning/milestones/. Run after /rcode-complete-milestone to keep .planning/ tidy. Shows a dry-run summary before moving anything."
+argument-hint: "[--dry-run]"
+allowed-tools: Read, Write, Bash, Glob, Grep, AskUserQuestion, Agent
+---
+
+<objective>
+Execute cleanup workflow
+</objective>
+
+<execution_context>
+@.rcode/workflows/cleanup.md
+</execution_context>
+
+<process>
+Execute the cleanup workflow from @.rcode/workflows/cleanup.md end-to-end.
+</process>

package/rcode/commands/code-review.md

@@ -0,0 +1,14 @@
+---
+name: rcode-review
+description: Review source files for bugs, security issues, and code quality problems.
+argument-hint: "<phase> [--depth=quick|standard|deep] [--files=file1,file2,...] [--karpathy] [--attack] [--edge-cases]"
+allowed-tools:
+  - Read
+  - Grep
+  - Glob
+  - Bash
+  - Task
+  - AskUserQuestion
+---
+
+@.rcode/workflows/code-review.md

package/rcode/commands/complete-milestone.md

@@ -0,0 +1,12 @@
+---
+name: rcode-complete-milestone
+description: Archive and reset — move completed milestone to archive and prepare for next cycle
+argument-hint: "[--archive-path=PATH]"
+allowed-tools:
+  - Read
+  - Bash
+  - Write
+  - AskUserQuestion
+---
+
+@.rcode/workflows/complete-milestone.md

package/rcode/commands/config.md

@@ -0,0 +1,8 @@
+---
+name: rcode-config
+description: Alias for /rcode-settings — view or edit rcode config (language, mode, model profile, workflow gates, git strategy)
+argument-hint: "[show | get <key> | set <key> <value>]"
+allowed-tools: Read, Write, Bash, AskUserQuestion
+---
+
+@.rcode/workflows/settings.md

package/rcode/commands/correct-course.md

@@ -0,0 +1,8 @@
+---
+name: rcode-correct-course
+description: Load original PRD/architecture, compare to current codebase. Classify deviation (scope drift / wrong architecture / missing AC / tech debt). Produce ordered remediation plan + updated story file.
+argument-hint: "[--prd <path>] [--architecture <path>]"
+allowed-tools: Read, Glob, Grep, Bash, Agent
+---
+
+@.rcode/workflows/correct-course.md

package/rcode/commands/council.md

@@ -0,0 +1,25 @@
+---
+name: rcode-council
+description: Convene the rcode majlis — spawns 3-5 specialist subagents in parallel to answer a strategic question. Agents are picked by keyword scoring.
+argument-hint: "<question> [--full] [--agents=a,b,c] [--explain]"
+allowed-tools:
+  - Read
+  - Grep
+  - Glob
+  - Bash
+  - Task
+  - AskUserQuestion
+---
+
+<objective>
+Convene 3-5 specialist agents in parallel to answer a strategic question. Agents picked by keyword scoring, responses synthesized into actionable decision.
+</objective>
+
+<execution_context>
+@.rcode/workflows/council.md
+</execution_context>
+
+<process>
+Execute the council workflow from @.rcode/workflows/council.md end-to-end.
+Use rcode-tools.cjs classify-question and select-panel to pick agents. Spawn agents in parallel. Synthesize responses.
+</process>

package/rcode/commands/create-architecture.md

@@ -0,0 +1,18 @@
+---
+name: rcode-create-architecture
+description: "Write an Architecture Decision Record (ADR) or system design document that informs implementation."
+argument-hint: "[topic]"
+allowed-tools: Read, Write, Bash, Glob, Grep, AskUserQuestion, Agent
+---
+
+<objective>
+Execute create-architecture workflow
+</objective>
+
+<execution_context>
+@.rcode/workflows/create-architecture.md
+</execution_context>
+
+<process>
+Execute the create-architecture workflow from @.rcode/workflows/create-architecture.md end-to-end.
+</process>

package/rcode/commands/create-epics-and-stories.md

@@ -0,0 +1,8 @@
+---
+name: rcode-create-epics-and-stories
+description: Parse a PRD or project document to generate numbered epic files in .planning/epics/. Each epic contains user stories with acceptance criteria and development notes.
+argument-hint: "<prd-path|project-path> [--prefix <name>]"
+allowed-tools: Read, Write, Glob, Grep, Bash, Agent
+---
+
+@.rcode/workflows/create-epics-and-stories.md

package/rcode/commands/create-prd.md

@@ -0,0 +1,18 @@
+---
+name: rcode-create-prd
+description: "Create a Product Requirements Document from scratch through guided facilitation. Use when starting a new project or feature that needs a PRD before architecture or phases."
+argument-hint: "[topic]"
+allowed-tools: Read, Write, Bash, Glob, Grep, AskUserQuestion
+---
+
+<objective>
+Execute create-prd workflow
+</objective>
+
+<execution_context>
+@.rcode/workflows/create-prd.md
+</execution_context>
+
+<process>
+Execute the create-prd workflow from @.rcode/workflows/create-prd.md end-to-end.
+</process>