Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/gf-labs/ramp/reviewgit clone --depth 1 https://github.com/gf-labs/rampWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/gf-labs/ramp/review)<a href="https://agentmods.dev/commands/gf-labs/ramp/review"><img src="https://agentmods.dev/badge/commands/gf-labs/ramp/review.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00011 | $0.02529 |
| Opus 5 | $0.00005 | $0.01264 |
| Sonnet 5 | $0.00002 | $0.00506 |
| Haiku 4.5 | $0.00001 | $0.00253 |
Grade B, and why
review scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
!`FIRST=$(echo "$ARGUMENTS" | awk '{print tolower($1)}'); if [ -n "$FIRST" ] && { [ -f "$HOME/.claude/ramp/schemas/$FIRST.md" ] || [ -f ".claude/knowledge-graphs/schemas/$FIRST.md" ]; }; then TOPIC="$FIRST"; else TOPIC=" How it starts
The opening of the file, as written. The whole thing — 154 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Context
Requested topic: $ARGUMENTS
Active topic (first word if it matches a known topic, otherwise "claude-code"):
!FIRST=$(echo "$ARGUMENTS" | awk '{print tolower($1)}'); if [ -n "$FIRST" ] && { [ -f "$HOME/.claude/ramp/schemas/$FIRST.md" ] || [ -f ".claude/knowledge-graphs/schemas/$FIRST.md" ]; }; then echo "$FIRST"; else echo "claude-code"; fi
Today's date: !date +%Y-%m-%d
Due nodes (kernel-computed SR queue for the active topic — the same rule as every due count ramp shows):
!FIRST=$(echo "$ARGUMENTS" | awk '{print tolower($1)}'); if [ -n "$FIRST" ] && { [ -f "$HOME/.claude/ramp/schemas/$FIRST.md" ] || [ -f ".claude/knowledge-graphs/schemas/$FIRST.md" ]; }; then TOPIC="$FIRST"; else TOPIC="claude-code"; fi; python3 "$CLAUDE_PLUGIN_ROOT/ramp_core.py" due "$TOPIC" 2>/dev/null || echo "DUE_UNAVAILABLE"
First-run signal (zero started topics ⇒ redirect a newcomer):
!python3 "$CLAUDE_PLUGIN_ROOT/ramp_core.py" catalog 2>/dev/null | python3 -c "import sys,json; c=json.load(sys.stdin); print('FIRST_RUN' if not any(t['started'] for t in c) else 'HAS_GRAPHS')" 2>/dev/null || echo "HAS_GRAPHS"
Knowledge graph contents (for active topic):
!FIRST=$(echo "$ARGUMENTS" | awk '{print tolower($1)}'); if [ -n "$FIRST" ] && { [ -f "$HOME/.claude/ramp/schemas/$FIRST.md" ] || [ -f ".claude/knowledge-graphs/schemas/$FIRST.md" ]; }; then TOPIC="$FIRST"; else TOPIC="claude-code"; fi; cat ~/.claude/ramp/graphs/$TOPIC.md 2>/dev/null || echo "NO_TREE_FILE:$TOPIC"
All topics — due counts (which other topics have due nodes):
!python3 "$CLAUDE_PLUGIN_ROOT/ramp_core.py" catalog 2>/dev/null | python3 -c "import sys,json; rows=[f\"{t['name']}: {t['summary']['due']} due\" for t in json.load(sys.stdin) if t.get('started') and t.get('summary') and t['summary']['due']>0]; print('\n'.join(rows) if rows else 'none')" 2>/dev/null || echo "none"
Your role
Empty-state redirect (check first): If the First-run signal (auto-collected context) is FIRST_RUN — no graphs exist yet — do not render an empty review. This takes precedence over the NO_TREE_FILE / "nothing due" handling below; short-circuit and say:
Nothing to review yet — you haven't started a topic. Run
/ramp:helpto get oriented, or/ramp:up <topic>to begin. See all topics with/ramp:list.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 5d ago First seen · 154 lines · 11 tokens per session scan B 97f91ce909be
review is a command published in the GitHub repository gf-labs/ramp (2 stars, last pushed 1mo ago), licensed MIT. It adds 11 tokens to every session and 2,529 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
design-tutorial
Interactive guided tour of Naksha — learn commands through real exercises, discover workflows, and get oriented in under 10 minutes.
practice-coding
Untimed guided coding practice: you write code in a seeded solution file in your own editor; a warm interviewer hints via a ladder, runs your code, and teaches inline. Summarized, not scored.
coding-drill
Rapid-fire reps — pattern recognition, complexity-of-this-snippet, template recall — flashcard cadence; prioritizes your weak patterns.
coding-import
Import a LeetCode problem into your local library via WebFetch — no MCP, no login. Becomes a normal markdown problem you can practice or mock.
alt
Import an Exam Radar (OPTIMETA Alt plugin) export and fold its lecture-emphasis exam signal into the course index — radar.md, a lecture-emphasis column on coverage.md, and a gold-zone weakmap.
blind
Strategy-level blind drill on a known HW or example problem. User describes approach in prose (no math typing); Claude verifies against solution then saves clean reference to derivations/.