aedt-bias-audit
HR-AI / AEDT bias audit. Invokes hr-ai-reviewer to assess NYC LL 144, EEOC, Illinois AIVIA, Colorado SB 205, EU AI Act Annex III applicability and produce…
Cost & capacity health — LLM router savings, run-rate, cost-per-deploy, ROI per shipped feature, WoW/MoM delta. Pairs with /digest (delivery+DORA) and /burn (reliability).
> /plugin marketplace add avelikiy/great_cto > /plugin install great_cto@great-cto
How it fires
How this command gets triggered: by you, by Claude, or both.
/costContext preview
What this command does when you run it.
Cost & capacity health — LLM router savings, run-rate, cost-per-deploy, ROI per shipped feature, WoW/MoM delta. Pairs with /digest (delivery+DORA) and /burn (reliability).
description: "Cost & capacity health — LLM router savings, run-rate, cost-per-deploy, ROI per shipped feature, WoW/MoM delta. Pairs with /digest (delivery+DORA) and /burn (reliability)." argument-hint: "[period_days] | feature <slug> | agent <name> — default: /cost 30. Examples: /cost 7 | /cost feature stripe-subscriptions | /cost agent architect" user-invocable: true allowed-tools: Read, Bash, Glob, Grep model: haiku
You are the Cost & Capacity aggregator. Multiple modes:
1. **`/cost [days]` — Aggregate health** (default mode)
2. **`/cost feature <slug>`** — ROI per shipped feature (NEW in v2.3.0)
3. **`/cost agent <name>`** — Per-agent cost (NEW in v2.3.0)
source .great_cto/env.sh 2>/dev/null || export PATH="/opt/homebrew/bin:$HOME/.local/bin:/usr/local/bin:$PATH"
# Detect mode from first arg
case "${1:-}" in
feature)
SLUG="${2:-}"
[ -z "$SLUG" ] && { echo "Usage: /cost feature <slug>"; exit 2; }
# → jump to "Feature mode" section below
MODE=feature
;;
agent)
AGENT="${2:-}"
[ -z "$AGENT" ] && { echo "Usage: /cost agent <name>"; exit 2; }
# → jump to "Agent mode" section below
MODE=agent
;;
*)
MODE=aggregate
PERIOD=${1:-30}
case "$PERIOD" in ''|*[!0-9]*) echo "Usage: /cost [period_days] | feature <slug> | agent <name>"; exit 2 ;; esac
;;
esac
COST_LOG=.great_cto/cost-history.log
DEPLOYS_LOG=.great_cto/deploys.logCompute total cost of shipping a feature, broken down by agent + ROI vs human equivalent.
if [ "$MODE" = "feature" ]; then
echo "## Cost: $SLUG"
echo ""
if [ ! -f "$COST_LOG" ]; then
echo "_No cost history yet — feature ROI requires at least one shipped feature._"
exit 0
fi
# Filter cost-history.log entries tagged with this feature slug
# Format: <timestamp> agent=<name> feature=<slug> cost_usd=<n> [other tags]
ENTRIES=$(grep -E "feature=$SLUG\b" "$COST_LOG" 2>/dev/null)
if [ -z "$ENTRIES" ]; then
echo "_No cost entries tagged with feature=$SLUG. Either:_"
echo "1. Feature not yet implemented (run \`/start \"feature description\"\`)"
echo "2. Feature uses different slug — check \`docs/architecture/ARCH-*.md\` filenames"
echo ""
echo "Available features in cost log:"
grep -oE "feature=[^ ]+" "$COST_LOG" | sort -u | head -10
exit 0
fi
# Aggregate by agent
echo "### Per-agent breakdown"
echo ""
echo "| Agent | Invocations | Cost | Avg/inv |"
echo "|-------|------------:|-----:|--------:|"
echo "$ENTRIES" | awk '
{
for (i=1;i<=NF;i++) {
if ($i ~ /^agent=/) { gsub(/agent=/, "", $i); agent = $i }
if ($i ~ /^cost[-_]?usd[=:]/) { gsub(/cost[-_]?usd[=:]/, "", $i); cost = $i+0 }
}
sum[agent] += cost
count[agent]++
}
END {
for (a in sum) printf "| %s | %d | $%.2f | $%.2f |\n", a, count[a], sum[a], sum[a]/count[a]
}
' | sort
TOTAL=$(echo "$ENTRIES" | awk '
{ for (i=1;i<=NF;i++) if ($i ~ /^cost[-_]?usd[=:]/) { gsub(/cost[-_]?usd[=:]/, "", $i); sum += $i+0 } }
END { printf "%.2f", sum }
')
echo "| **Total** | | **\$$TOTAL** | |"
echo ""
# Human-equivalent comparison
ARCHETYPE=$(grep -E "^archetype:|^primary:" .great_cto/PROJECT.md 2>/dev/null | head -1 | awk '{print $2}')
# Default: assume 12 hours human equivalent for standard feature
HUMAN_HOURS=${HUMAN_HOURS:-12}
HUMAN_RATE=${HUMAN_RATE:-150}
HUMAN_COST=$(echo "scale=0; $HUMAN_HOURS * $HUMAN_RATE" | bc)
ROI=$(echo "scale=1; $HUMAN_COST / $TOTAL" | bc)
echo "### vs Human equivalent"
echo ""
echo "- Estimated effort: ${HUMAN_HOURS}h × \$${HUMAN_RATE}/hr = **\$${HUMAN_COST}**"
echo "- AI cost: **\$${TOTAL}**"
echo "- **ROI: ${ROI}x**"
echo ""
echo "_Override estimates: \`HUMAN_HOURS=20 HUMAN_RATE=200 /cost feature $SLUG\`_"
# Comparison to mean for archetype
if [ -n "$ARCHETYPE" ]; then
echo ""
echo "### Compared to other features in archetype=$ARCHETYPE"
grep "feature=" "$COST_LOG" | grep -v "feature=$SLUG" | awk '
{
for (i=1;i<=NF;i++) {
if ($i ~ /^feature=/) { gsub(/feature=/, "", $i); feat = $i }
if ($i ~ /^cost[-_]?usd[=:]/) { gsub(/cost[-_]?usd[=:]/, "", $i); cost = $i+0 }
}
sum[feat] += cost
}
END {
for (f in sum) printf "- %s: $%.2f\n", f, sum[f]
}
' | sort -t '$' -k2 -n | head -5
fi
exit 0
fiQuick per-agent cost summary (lighter than /agent-review):
if [ "$MODE" = "agent" ]; then
echo "## Cost: $AGENT"
echo ""
if [ ! -f "$COST_LOG" ]; then
echo "_No cost history yet._"
exit 0
fi
ENTRIES=$(grep -E "agent=$AGENT\b" "$COST_LOG" 2>/dev/null)
COUNT=$(echo "$ENTRIES" | grep -c .)
if [ "$COUNT" = "0" ]; then
echo "_No cost entries for agent=$AGENT._"
exit 0
fi
TOTAL=$(echo "$ENTRIES" | awk '{ for (i=1;i<=NF;i++) if ($i ~ /^cost[-_]?usd[=:]/) { gsub(/cost[-_]?usd[=:]/, "", $i); sum += $i+0 } } END { printf "%.2f", sum }')
AVG=$(echo "scale=2; $TOTAL / $COUNT" | bc)
echo "- Total invocations: $COUNT"
echo "- Total cost: \$$TOTAL"
echo "- Avg cost/invocation: \$$AVG"
echo ""
echo "_For full performance review: \`/agent-review $AGENT\`_"
exit 0
fiYou are the Cost & Capacity aggregator. Two parts:
1. **LLM router savings** — measured Kimi-vs-Sonnet differential from `.great_cto/llm-router-usage.log`. This is the data behind the README
You already have the agent. This is everything around it. great_cto runs Claude Code as a pipeline of 70 specialist agents — an independent model checks each stage before the next builds on it, spending caps refuse rather than warn, and three decisions stay yours: what gets built, how, and whether it ships.
Repo: avelikiy/great_cto
HR-AI / AEDT bias audit. Invokes hr-ai-reviewer to assess NYC LL 144, EEOC, Illinois AIVIA, Colorado SB 205, EU AI Act Annex III applicability and produce…
Gracefully retire an LLM agent from the workforce. Archives prompt, removes from sync list, keeps verdicts for audit. Like firing a human — but reversible.
Performance review for an LLM agent (or all agents). Verdicts breakdown, cost analysis, top failure modes, prompt-tuning suggestions. Like a human '1:1' but…
API platform contract review. Invokes api-platform-reviewer to audit rate-limit design, OAuth scope hygiene, webhook signing, idempotency, Sunset/deprecation,…
Audit an existing codebase. Detects stack, finds gaps, creates tasks, generates PROJECT.md.
Open the great_cto admin board at http://localhost:3141 (Kanban, cost, pipeline, inbox, memory). Starts it in background if not running.