π The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated
TypeScript agentic-aiagentic-frameworkagentic-workflowagentsai-agentsai-assistantai-skillsautonomous-agentsclaudecodexdsh-pluginharnessmcp-servermulti-agentmulti-agent-systemsnpmskillsswarmswarm-intelligencetypescript
feat(cost-tracker): P3 β auto-emit hooks_model-outcome (v0.7.0)
Reuven333e6338d9c3
diff --git a/plugins/ruflo-cost-tracker/.claude-plugin/plugin.json b/plugins/ruflo-cost-tracker/.claude-plugin/plugin.jsonindex 74f07e19f8..defb26f01b 100644--- a/plugins/ruflo-cost-tracker/.claude-plugin/plugin.json+++ b/plugins/ruflo-cost-tracker/.claude-plugin/plugin.json@@ -1,7 +1,7 @@{"name": "ruflo-cost-tracker","description": "Token usage tracking, model cost attribution per agent, budget alerts, and optimization recommendations β uses memory_* (namespace-routed) for cost-tracking and cost-patterns; pairs with federation budget circuit breaker (ADR-097)",- "version": "0.6.0",+ "version": "0.7.0","author": {"name": "ruvnet","url": "https://github.com/ruvnet"@@ -24,6 +24,7 @@"benchmarking","verified","telemetry",- "budget"+ "budget",+ "model-outcome"]}diff --git a/plugins/ruflo-cost-tracker/README.md b/plugins/ruflo-cost-tracker/README.mdindex 20f76f3e73..502358f0e4 100644--- a/plugins/ruflo-cost-tracker/README.md+++ b/plugins/ruflo-cost-tracker/README.md@@ -41,7 +41,8 @@ cost budget check [--period ...] # Compute utilization + alert; exit 1cost report [--period today|week|month] # Cost report (with By-tier block, reads measured booster data)cost breakdown [--by agent|model|task] # Detailed breakdown by dimensioncost budget set <amount> # Set budget limit in USD-cost optimize # Analyze usage and suggest savings (+ hooks_model-outcome feedback)+cost optimize # Analyze usage and suggest savings (+ auto-emits hooks_model-outcome via outcome.mjs)+cost outcome <task> <model> <outcome> # Emit hooks_model-outcome (success|escalated|failure) so the router learnscost benchmark [--llm] [--anthropic] # Run measured benchmark β booster + optional Gemini/Sonnet/Opus baselinescost workers # Inspect optimize + benchmark loop-workers consumedcost history # Show cost tracking over time@@ -130,7 +131,7 @@ Reserved namespaces (`pattern`, `claude-memories`, `default`) MUST NOT be shadow```bashbash plugins/ruflo-cost-tracker/scripts/smoke.sh-# Expected: "33 passed, 0 failed"+# Expected: "36 passed, 0 failed"```## Architecture Decisionsdiff --git a/plugins/ruflo-cost-tracker/commands/ruflo-cost.md b/plugins/ruflo-cost-tracker/commands/ruflo-cost.mdindex 865b17b92d..983b8b89c8 100644--- a/plugins/ruflo-cost-tracker/commands/ruflo-cost.md+++ b/plugins/ruflo-cost-tracker/commands/ruflo-cost.md@@ -49,6 +49,12 @@ Cost tracking commands:2. Print: total cost, per-model and per-tier breakdown, persisted memory key3. Sets the `cost-tracking` namespace record at key `session-<sessionId>` (consumed by `cost-report` step 1)+**`cost outcome <task> <model> <outcome>`** -- Emit a `hooks_model-outcome` event so the router learns from applied recommendations. Auto-wired into `cost-optimize` step 8.+1. Validates `outcome β {success, escalated, failure}`+2. Runs `node plugins/ruflo-cost-tracker/scripts/outcome.mjs "<task>" <model> <outcome>`+3. The script wraps `npx @claude-flow/cli hooks model-outcome -t ... -m ... -o ...` with explicit-argv spawnSync so quoting is safe+4. Without this, the router doesn't learn from cost-optimize recommendations and the Tier 1 bypass rate doesn't tighten over time+**`cost benchmark [--llm] [--anthropic]`** -- Run the corpus benchmark to verify booster claims with measured numbers.1. Without flags: booster-only (free, ~85 ms wall-time, no API keys needed)2. `--llm`: also run Gemini 2.0 Flash baseline (uses GCP `GOOGLE_AI_API_KEY` secret)diff --git a/plugins/ruflo-cost-tracker/scripts/outcome.mjs b/plugins/ruflo-cost-tracker/scripts/outcome.mjsnew file mode 100755index 0000000000..cb17ccab74--- /dev/null+++ b/plugins/ruflo-cost-tracker/scripts/outcome.mjs@@ -0,0 +1,43 @@+#!/usr/bin/env node+// cost-outcome β thin wrapper around `hooks model-outcome` so cost-optimize+// recommendations can be emitted as pre-formatted shell commands. The agent+// (or human) runs this once per applied recommendation and the router learns.+//+// Usage:+// node scripts/outcome.mjs <task-description> <model> <outcome>+// outcome β { success, escalated, failure }+//+// Example:+// node scripts/outcome.mjs "format imports" haiku success+// node scripts/outcome.mjs "design auth flow" sonnet escalated+//+// Why a wrapper: `hooks model-outcome` takes `-t -m -o` and quoting through+// shell pipelines is error-prone. This script uses spawnSync with explicit+// argv so a task description with quotes/newlines is preserved.++import { spawnSync } from 'node:child_process';++const ALLOWED = new Set(['success', 'escalated', 'failure']);++function main() {+ const [task, model, outcome] = process.argv.slice(2);+ if (!task || !model || !outcome) {+ console.error('usage: outcome.mjs <task-description> <model> <outcome>');+ console.error(' outcome β {success, escalated, failure}');+ process.exit(2);+ }+ if (!ALLOWED.has(outcome)) {+ console.error(`invalid outcome '${outcome}' β must be one of: success, escalated, failure`);+ process.exit(2);+ }+ const r = spawnSync('npx', [+ '@claude-flow/cli@latest', 'hooks', 'model-outcome',+ '-t', task, '-m', model, '-o', outcome,+ ], { stdio: 'inherit' });+ if (r.status !== 0) {+ console.error(`emit failed (exit ${r.status})`);+ process.exit(r.status || 1);+ }+}++main();diff --git a/plugins/ruflo-cost-tracker/scripts/smoke.sh b/plugins/ruflo-cost-tracker/scripts/smoke.shindex 95a79d65ed..f6d31d7c33 100755--- a/plugins/ruflo-cost-tracker/scripts/smoke.sh+++ b/plugins/ruflo-cost-tracker/scripts/smoke.sh@@ -8,10 +8,10 @@ step() { printf "β %s ... " "$1"; }ok() { printf "PASS\n"; PASS=$((PASS+1)); }bad() { printf "FAIL: %s\n" "$1"; FAIL=$((FAIL+1)); }-step "1. plugin.json declares 0.6.0 with new keywords"+step "1. plugin.json declares 0.7.0 with new keywords"v=$(grep -E '"version"' "$ROOT/.claude-plugin/plugin.json" | grep -oE '[0-9]+\.[0-9]+\.[0-9]+' | head -1)-if [[ "$v" != "0.6.0" ]]; then- bad "expected 0.6.0, got '$v'"+if [[ "$v" != "0.7.0" ]]; then+ bad "expected 0.7.0, got '$v'"elsemiss=""for k in namespace-routing mcp agentic-flow agent-booster tier1-routing model-routing benchmarking verified telemetry budget; do@@ -271,5 +271,24 @@ grep -q "cost budget check" "$F" || miss="$miss check"grep -qE "50/75/90/100|alert ladder|HARD_STOP" "$F" || miss="$miss alert-ladder"[[ -z "$miss" ]] && ok || bad "$miss"+step "34. cost-optimize step 8 wires auto-emit via outcome.mjs"+F="$ROOT/skills/cost-optimize/SKILL.md"+grep -q "outcome\.mjs" "$F" && grep -qE "success|escalated|failure" "$F" \+ && ok || bad "step 8 not wired to outcome.mjs"++step "35. outcome.mjs harness present + parses + validates outcomes"+F="$ROOT/scripts/outcome.mjs"+miss=""+[[ -x "$F" ]] || miss="$miss not-executable"+node --check "$F" 2>/dev/null || miss="$miss syntax-error"+grep -q "spawnSync" "$F" || miss="$miss no-spawnSync"+grep -q "hooks.*model-outcome" "$F" || miss="$miss no-hooks-call"+grep -qE "success.*escalated.*failure|ALLOWED" "$F" || miss="$miss no-validation"+[[ -z "$miss" ]] && ok || bad "$miss"++step "36. ruflo-cost.md documents 'cost outcome' subcommand"+grep -q "cost outcome" "$ROOT/commands/ruflo-cost.md" \+ && ok || bad "missing"+printf "\n%s passed, %s failed\n" "$PASS" "$FAIL"[[ $FAIL -eq 0 ]] || exit 1diff --git a/plugins/ruflo-cost-tracker/skills/cost-optimize/SKILL.md b/plugins/ruflo-cost-tracker/skills/cost-optimize/SKILL.mdindex c6cf0491bc..74f13e3d89 100644--- a/plugins/ruflo-cost-tracker/skills/cost-optimize/SKILL.md+++ b/plugins/ruflo-cost-tracker/skills/cost-optimize/SKILL.md@@ -26,7 +26,17 @@ When costs are higher than expected or you want to proactively reduce spending.7. **Store the optimization pattern** -- two paths:- **Pattern store (typed, recommended)**: `mcp__claude-flow__agentdb_pattern-store` with `type: 'cost-optimization'`. Don't pass a `namespace` arg β ReasoningBank routes it; on bridge unavailability the fallback writes to the reserved `pattern` namespace with `controller: 'memory-store-fallback'` (see ruflo-agentdb ADR-0001).- **Plain store (namespace-routable)**: `mcp__claude-flow__memory_store --namespace cost-patterns` β this DOES respect the `cost-patterns` namespace because `memory_*` is namespace-routed.-8. **Close the routing feedback loop** -- when a downgrade recommendation is *applied* (or its application is logged elsewhere in this session), call `mcp__claude-flow__hooks_model-outcome` with `{task, fromModel, toModel, outcome: 'success'|'escalated'|'failure'}`. This is the typed equivalent of the legacy `routing-outcomes` namespace (see ruflo-intelligence ADR-0001 Β§"Neutral"). Without this signal the router does not learn from cost-tracker's recommendations and the booster bypass rate (see `cost-booster-route` skill) does not improve over time.+8. **Close the routing feedback loop β auto-emit `hooks_model-outcome`** -- for each downgrade recommendation, format the outcome-emit command as part of the recommendation table so it can be run directly:++ ```bash+ # success path (downgrade worked)+ node plugins/ruflo-cost-tracker/scripts/outcome.mjs "<task-description>" <model> success++ # escalated path (had to upgrade after downgrade attempt)+ node plugins/ruflo-cost-tracker/scripts/outcome.mjs "<task-description>" <model> escalated+ ```++ The script wraps `npx @claude-flow/cli hooks model-outcome -t ... -m ... -o ...` with explicit-argv `spawnSync` so quoting is safe. Without this signal the router does not learn from cost-tracker's recommendations and the booster bypass rate (see `cost-booster-route` skill) does not improve over time. This is the typed equivalent of the legacy `routing-outcomes` namespace (see ruflo-intelligence ADR-0001 Β§"Neutral").9. **Report** -- display: ranked recommendations with savings estimate, total potential savings, implementation priority (quick wins first), and any model-outcome events emitted in step 8## CLI alternative