Slash Commands
The Stellarion repo ships 14 ready-to-use slash commands for Claude Code and the matching set for VS Code Copilot Chat. Each one chains 2-5 Stellarion tool calls and produces a focused Markdown report — designed for tech leads, architects, and engineering managers.
Quick install
Claude Code
git clone https://github.com/stellarion-ai/stellarion.git /tmp/stellarion
mkdir -p ~/.claude/commands
cp /tmp/stellarion/examples/claude-code/commands/*.md ~/.claude/commands/
The commands now appear when you type / in Claude Code.
VS Code Copilot Chat
mkdir -p .github/prompts
cp /tmp/stellarion/examples/vscode-copilot/prompts/*.prompt.md .github/prompts/
The prompts now appear when you type / in Copilot Chat.
Per-project install
If you want the commands available only inside one repo, copy them into <repo>/.claude/commands/ instead of ~/.claude/commands/. Same for <repo>/.github/prompts/ for Copilot.
EDA-mode commands (3)
Require --mode eda and a Pro license / active trial. Compose find_eda_calls, get_flow_graph, and find_duplicate_constraints.
| Command | What it does |
|---|---|
/eda-flow-audit [scope] | Maps the EDA flow architecture: 18 stages with active counts, top members, cross-stage transitions, and stage gaps to flag. Identifies flow orchestrators (procs spanning ≥3 stages). |
/eda-drift-check [category] | Runs find_duplicate_constraints with onlyTrueConflicts: true and classifies each result as 🔴 likely bug / 🟡 parameterization candidate / 🟢 managed family fork. |
/eda-cross-family-clones | Finds _hw.tcl and Tcl clones across FPGA family directories (stratix10 / agilex / arria10) and proposes a unification plan: promote-to-common, parameterize, or keep forked. |
Architect / lead-developer commands (7)
Cover the recurring architect questions: ship-readiness, complexity, PR risk, refactor planning, dead code, API surface, coupling. Each includes EDA-aware sections that activate when the codebase has Quartus / Qsys / SDC content.
| Command | What it does |
|---|---|
/architecture-review | Multi-axis ship-readiness review (health, debt, drift, security, API, EDA flow + drift). Returns a TL;DR, top 3 risks, debt punch list, recommendations. |
/complexity-hotspots [N] | Top-N complexity hotspots (cyclomatic + cognitive) with refactor priority, test coverage, and an EDA category column showing each hotspot's dominant Quartus / Qsys family. |
/change-risk [base-ref] | PR-style risk assessment of the current branch's changes. Includes EDA blast-radius: cross-flow files and any new constraint conflicts the change introduces. |
/refactor-candidates | Surfaces refactor opportunities from duplicates, clusters, similar-impl matches, + EDA constraint drift. Classifies each as Type A-E with action recommendations. |
/dead-code-sweep [confidence] | Unused-code candidates with a topologically-sorted removal plan (delete leaves before parents to avoid spurious matches in the next pass). |
/api-surface-review [baseline-tag] | Pre-release API surface review with semver-bump recommendation. Also catalogs Qsys component declarations (set_module_property NAME …) as the EDA-mode equivalent of public symbols. |
/coupling-audit | Module coupling smells (stable hubs / coupling pain / cycles) + EDA flow choke points (procs spanning ≥3 flow stages). |
Executive reporting commands (4)
For VP Eng / Director / monthly review use. Outputs are tight (under 50-100 lines), KPI-driven, jargon-free, and end with concrete eng-week-sized recommendations.
| Command | What it does |
|---|---|
/complexity-executive-report [scope] | One-page executive view: KPI table with grade, complexity-distribution histogram, top 5 risks, and "what it would take to lift the grade" expressed in eng-weeks. Includes EDA constraint conflicts and flow-stage-active KPIs. |
/complexity-trend-report [baseline] | Compares HEAD against a baseline (branch / tag / SHA). Reports direction (🟢 improving / 🟡 stable / 🔴 regressing), headline deltas (incl. EDA drift count), hotspots gained / resolved, merge verdict. |
/refactor-roi-report [N] | Prioritized refactor backlog with explicit ROI = impact ÷ effort. Categorizes items as 🔥 Quick win / 💪 High value / 🧱 Big rock / ⚙️ Hygiene / ❄️ Skip. Includes an eda-drift item type with its own ROI scoring formula. Ends with a sprint-shaped commitment ("2 eng-weeks closes 47% of hotspots"). |
/quality-scorecard | Multi-dimensional 7-axis scorecard: 🩺 health · 🧠 complexity · 💰 debt · 🛡️ security · 🧪 tests · 🔗 coupling · 🔧 EDA flow & drift. Each scored 0-100 with letter grade A-F. Composite + 3-step movement plan. |
Anatomy of a slash command
The Claude Code .md files have YAML frontmatter and a prompt body:
---
description: One-line summary shown in the slash-command picker
argument-hint: [optional placeholder for args]
---
The actual prompt sent to Claude when this command is invoked.
$ARGUMENTS / $1 / $2 are replaced with whatever the user types.
The Copilot .prompt.md mirrors use a slightly different frontmatter:
---
mode: 'agent'
description: '...'
tools: ['stellarion_xxx', 'stellarion_yyy']
---
Same prompt body as the Claude Code version.
${input:scope} replaces $ARGUMENTS.
The prompt bodies are kept identical so the two flavors stay in lockstep — only frontmatter and arg syntax differ.
Customizing
To override a project default with your own version:
# Copy a command, edit it, drop it in your user-scope directory.
cp ~/.claude/commands/architecture-review.md ~/.claude/commands/architecture-review.md.bak
$EDITOR ~/.claude/commands/architecture-review.md
User-scope commands shadow project-scope ones with the same name.
See also
- Tools Overview — every Stellarion tool the slash commands compose
- Claude Code integration — install path and EDA mode
- Cursor integration — same MCP server, different host