From fellowship
Interactive tour of fellowship. Detects project state and recommends a starting command. Accepts a task description to classify and recommend.
How this command is triggered — by the user, by Claude, or both
Slash command
/fellowship:help [task description | --phase plan|build|review|ship]haikuThis command is limited to the following tools:
The summary Claude sees in its command listing — used to decide when to auto-load this command
# /fellowship:help Interactive, context-aware tour of the **fellowship** plugin. Scans the current project for artifacts and state files, then recommends the right command to run next. Three modes: - **No argument** — full lifecycle tour - **`--phase <name>`** — scoped tour of one phase (`plan`, `build`, `review`, `ship`) - **Task description** — classify the task and **print a recommended command**. This command does NOT auto-run the recommendation — the user runs it themselves. This command is read-only. Never modifies files. --- ## ON STARTUP — Scan Project State Run all scans in p...
Interactive, context-aware tour of the fellowship plugin. Scans the current project for artifacts and state files, then recommends the right command to run next.
Three modes:
--phase <name> — scoped tour of one phase (plan, build, review, ship)This command is read-only. Never modifies files.
Run all scans in parallel, then choose a greeting and tour path.
Deferred tool note: Before calling
AskUserQuestionfor the first time, callToolSearchwith query"select:AskUserQuestion"to load the tool schema.
Glob: .skilmarillion/projects/*/PRD.md
Glob: .skilmarillion/projects/*/ROADMAP.md
Glob: .skilmarillion/projects/*/specs/SPEC-*.md
Glob: .skilmarillion/projects/*/adrs/*.md
Glob: .skilmarillion/projects/*/reviews/*.md
Store counts as {prds}, {roadmaps}, {specs}, {adrs}, {review_reports}.
Glob: .skilmarillion/projects/*/PROJECT-STATE.yaml
For each file, read it and detect which sections are present (plan:, impl:, review:). Store as {active_states}.
If $ARGUMENTS contains a task description (not a flag), classify intent by keyword match:
| Signal | Recommended Command |
|---|---|
| plan, spec, prd, roadmap, requirements, acceptance criteria, user story, scope, define | /fellowship:plan |
| architecture, adr, api design, schema, database, diagram, data model, endpoint, openapi, erd | /fellowship:plan --arch <type> |
| build, implement, tdd, write code, fix bug, debug, refactor, add feature, create endpoint | /fellowship:build |
| review, audit, security, accessibility, a11y, quality, lint, vulnerability, ai slop, clean | /fellowship:review |
| commit, pr, pull request, ship, merge, conventional commit | /fellowship:ship |
If ambiguous or none match, ask AskUserQuestion:
"Which phase fits your task?"
- Plan — define what to build (
/fellowship:plan)- Build — implement with TDD (
/fellowship:build)- Review — audit code quality (
/fellowship:review)- Ship — commit + PR (
/fellowship:ship)
Output format (recommendation only, no delegation):
Recommendation: "{task}" →
/fellowship:{command}Run:
/fellowship:{command} {task}(not auto-executing — your call)
Stop. Do NOT invoke the target command.
--phase <name> flagJump straight to the matching phase section in the tour below. No greeting, no summary — just that one phase.
Display greeting based on scan results, then walk all phases.
Choose based on scan results:
Fresh project (no artifacts, no state):
Welcome to fellowship. No specs, PRDs, or active workflows yet. Let me show you around.
Existing artifacts:
Welcome back. Found {specs} spec(s), {prds} PRD(s), {review_reports} review report(s). Here's the lay of the land.
Active workflows:
Welcome back. You have active work in progress. Here's where you stand.
Then show the 7-command map:
init → plan → build → review → ship
status | help
Ask:
"Full tour, or jump to a phase?"
- Full tour
- Jump to: plan | build | review | ship
Walk through each phase one at a time. After each, ask whether to continue.
/fellowship:plan (Council)Plan — spec-driven planning
Turns a task description into a testable spec and TDD plan before any code is written.
Stages (select via flags):
--prd— client-shareable Product Requirements Document--roadmap <prd-path>— decompose an approved PRD into ordered milestones--specify <roadmap-path>— generate SPEC files from a roadmap (parallel agents)--migrate <legacy> <target>— prioritized migration plan--validate <path>— score a PRD/spec/plan (0-100; PASS at ≥85 for all types)--arch <adr|api|schema|diagram>— design-session artifactsDefault (no flag): runs the guided PRD → ROADMAP → specs flow end to end.
Artifacts:
.skilmarillion/projects/{slug}/{PRD.md | ROADMAP.md | specs/ | adrs/ | api/ | schema/ | diagrams/}
Ask:
"Continue to Build, jump elsewhere, or end the tour?"
/fellowship:build (Journey)Build — slice-by-slice TDD execution
Takes a spec and drives RED → GREEN → REFACTOR, one slice at a time. Resumes from
PROJECT-STATE.yamlwhen state is found.Modes:
- (default) — TDD loop
--debug <issue>— structured debugging: reproduce → isolate → root cause → fix--refactor <target>— phase-gated refactoring with tests between stepsEscalation: after 3 failed attempts on a slice, diagnostic step → modified approach, sub-slice split, or ACCEPT_WITH_DEBT.
Ask:
"Continue to Review, jump elsewhere, or end the tour?"
/fellowship:review (Rivendell)Review — parallel quality gates, findings only
Spawns specialist reviewers in parallel, deduplicates findings, sorts by impact-to-effort. Never modifies code.
Filters:
- (default) — full review (code quality + security + a11y)
--security— security only (>80% confidence threshold)--a11y— WCAG 2.1/2.2 (uses Playwright MCP when available)--clean— AI-generated noise detectionArtifact:
.skilmarillion/projects/{slug}/reviews/review-{target}.md
Ask:
"Continue to Ship, jump elsewhere, or end the tour?"
/fellowship:shipShip — conventional commit + PR
Generates a conventional commit from staged changes. Detects breaking changes, infers type and scope, asks for approval before committing.
Flags:
- (default) — commit only
--pr— push branch + open PR with AC traceability--breaking— force BREAKING CHANGE footer--scope <scope>— override auto-detected scope
Ask:
"Ready for a recommended starting point?"
Based on the project state scan, recommend the next move:
Active state files exist →
Recommendation: You have active work. Run
/fellowship:statusto see where it stands, then resume with the matching command.
PRDs exist, no specs for those features →
Recommendation: Run
/fellowship:plan --roadmap {prd-path}to decompose the PRD into milestones, then/fellowship:plan --specify {roadmap-path}.
Specs exist, no in-progress build →
Recommendation: Run
/fellowship:build {spec-path}to start TDD execution.
Fresh project, no artifacts →
Recommendation: Run
/fellowship:planto start the guided PRD → ROADMAP → specs flow, or/fellowship:plan --prd "[feature description]"to skip straight to a PRD.
Always end the tour with:
Other commands:
/fellowship:init— bootstrap.skilmarillion/projects/{slug}/scaffolding/fellowship:status— read-only dashboard of active work/fellowship:help [task]— classify a task and recommend a commandRun
/fellowship:help --phase <name>for a scoped tour of one phase.
/helpExplains Ralph Loop plugin for iterative AI development via repeated Claude prompts with file self-reference, and lists commands /ralph-loop to start loops and /cancel-ralph to cancel.
/helpDisplays categorized overview of Task Master AI commands including setup, tasks, status, analysis. Accepts arguments for specific command help.
/helpDisplays Hookify plugin help: hook events, markdown config files with regex patterns, creation via /hookify or manual files, subcommands, and examples for safe behaviors.
/helpDisplays full reference guide for Ouroboros commands and agents by rendering the dedicated SKILL.md file.
/helpDisplays the GSD help reference at the requested tier — one-line summary, one-page tour, full reference, or a single topic section.
/helpDisplays a complete table of all available SuperClaude (/sc) commands with their descriptions and functionality.
npx claudepluginhub trevoredris/skilmarillion