Files
foxhunt/.claude/helpers/guidance-hooks.sh
jgrusewski 12fdd18223 refactor: remove entire CPU training path — 5,307 lines of dead code
Deleted:
- DQN::compute_loss_internal (280 lines) — old Candle forward+loss
- DQN::train_step (55 lines) — old Candle training step
- DQN::compute_gradients (47 lines) — old gradient accumulation
- ComputeLossResult struct — only used by deleted functions
- RegimeConditionalDQN::train_step (65 lines) — old dispatch
- RegimeConditionalDQN::train_step_gpu_regime (100 lines) — old GPU path
- RegimeConditionalDQN::compute_gradients_gpu (130 lines) — old regime gradients
- RegimeConditionalDQN::compute_gradients (92 lines) — old dispatch
- DQNAgentType::train_step dispatch — dead
- DQNAgentType::compute_gradients dispatch — dead
- GpuDqnTrainer::upload_batch (71 lines) — old CPU→GPU upload
- train_step.rs (500 lines) — entire module including ensure_fused_ctx
- dqn_benchmark.rs — used old train_step
- examples.rs — used old train_step
- validation/adapters.rs (289 lines) — used old train_step
- dqn/trainable_adapter.rs — used old train_step
- gpu_smoketest.rs — tested old train_step
- Gradient accumulation path in training_loop.rs (144 lines)
- IQN d_h_s2().clone() → raw pointer (zero alloc)
- Causal intervention format! string alloc removed
- Dead HER relabel functions (320 lines)

Kept:
- ensure_fused_ctx logic inlined into training_loop.rs
- set_noise_sigma_scale re-added to RegimeConditionalDQN

Fixed:
- GpuReplayBuffer max_batch_size wired from batch_size parameter
  (was hardcoded 1024, blocking batch_size=8192)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-02 09:06:23 +02:00

103 lines
2.7 KiB
Bash
Executable File

#!/bin/bash
# Guidance Hooks for Claude Flow V3
# Provides context and routing for Claude Code operations
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
PROJECT_ROOT="$(cd "$SCRIPT_DIR/../.." && pwd)"
CACHE_DIR="$PROJECT_ROOT/.claude-flow"
# Ensure cache directory exists
mkdir -p "$CACHE_DIR" 2>/dev/null || true
# Color codes
CYAN='\033[0;36m'
GREEN='\033[0;32m'
YELLOW='\033[1;33m'
RED='\033[0;31m'
RESET='\033[0m'
DIM='\033[2m'
# Get command
COMMAND="${1:-help}"
shift || true
case "$COMMAND" in
pre-edit)
FILE_PATH="$1"
if [[ -n "$FILE_PATH" ]]; then
if [[ "$FILE_PATH" =~ (config|secret|credential|password|key|auth) ]]; then
echo -e "${YELLOW}[Guidance] Security-sensitive file${RESET}"
fi
if [[ "$FILE_PATH" =~ ^v3/ ]]; then
echo -e "${CYAN}[Guidance] V3 module - follow ADR guidelines${RESET}"
fi
fi
exit 0
;;
post-edit)
FILE_PATH="$1"
echo "$(date -Iseconds) edit $FILE_PATH" >> "$CACHE_DIR/edit-history.log" 2>/dev/null || true
exit 0
;;
pre-command)
COMMAND_STR="$1"
if [[ "$COMMAND_STR" =~ (rm -rf|sudo|chmod 777) ]]; then
echo -e "${RED}[Guidance] High-risk command${RESET}"
fi
exit 0
;;
route)
TASK="$1"
[[ -z "$TASK" ]] && exit 0
if [[ "$TASK" =~ (security|CVE|vulnerability) ]]; then
echo -e "${DIM}[Route] security-architect${RESET}"
elif [[ "$TASK" =~ (memory|AgentDB|HNSW|vector) ]]; then
echo -e "${DIM}[Route] memory-specialist${RESET}"
elif [[ "$TASK" =~ (performance|optimize|benchmark) ]]; then
echo -e "${DIM}[Route] performance-engineer${RESET}"
elif [[ "$TASK" =~ (test|TDD|spec) ]]; then
echo -e "${DIM}[Route] test-architect${RESET}"
fi
exit 0
;;
session-context)
cat << 'EOF'
## V3 Development Context
**Architecture**: Domain-Driven Design with 15 @claude-flow modules
**Priority**: Security-first (CVE-1, CVE-2, CVE-3 remediation)
**Performance Targets**:
- HNSW search: 150x-12,500x faster
- Flash Attention: 2.49x-7.47x speedup
- Memory: 50-75% reduction
**Active Patterns**:
- Use TDD London School (mock-first)
- Event sourcing for state changes
- agentic-flow@alpha as core foundation
- Bounded contexts with clear interfaces
**Code Quality Rules**:
- Files under 500 lines
- No hardcoded secrets
- Input validation at boundaries
- Typed interfaces for all public APIs
**Learned Patterns**: 17 available for reference
EOF
exit 0
;;
user-prompt)
exit 0
;;
*)
exit 0
;;
esac