24 projects
daedalus-ai
Daedalus — self-evolving multimodal coding assistant powered by the Hermes Deep Mind engine: persistent memory, failure immunity, sleep-time compute, causal world model, MoE expert routing, calibrated confidence, MCP client, and 24 LLM providers
fableforge-agent
FableForge Agent — a fork of Hermes Agent (NousResearch). A self-improving AI agent that creates skills from experience, improves them during use, and runs anywhere
closed-loop-training
ReMiT + RLPT closed-loop flywheel: continuous self-improving LLM training
hermes-ultimate
Self-evolving multimodal coding assistant: persistent memory, failure immunity, sleep-time compute, causal world model, MoE expert routing, calibrated confidence, MCP client, and 22 LLM providers
fableforge-anvil-agent
The open-source, self-verified coding agent. Generate → Execute → Verify → Recover.
fableforge-cli
Umbrella CLI for the FableForge agent ecosystem
fableforge
Unified CLI for the FableForge agent ecosystem. 21 projects, one command.
fable5-dataset
Load, preprocess, and manage the Fable5 agent trace datasets
fableforge-trace-compiler
Compile agent traces into distilled LoRA weights for specific coding behaviors
fableforge-cost-optimizer
Analyze token waste and optimize LLM routing for cost savings
fableforge-trajectory-distiller
Distill agent traces into training datasets across multiple formats
fableforge-bench-agent
HumanEval for tool use — a standardized benchmark for evaluating LLM tool-use capabilities
fableforge-agent-runtime
Persistent agent daemon — systemd for AI agents
fableforge-agent-skills
Skill registry, decomposition, and LoRA building for coding agents
fableforge-agent-telemetry
Datadog for AI agents — real-time observability, token tracking, cost estimation, error rates
fableforge-agent-profiler
Profile and classify agent behavior patterns from traces
agent-curriculum
Curriculum learning for agent training — difficulty-scored stages
agent-fuzzer
Adversarial scenario generation and testing for coding agents
fableforge-shell-whisperer
1.5B edge-native shell agent — natural language to shell commands, 50ms latency on phone/edge
fableforge-agent-swarm
Orchestrate micro-agent swarms using Markov transition matrices as handoff patterns
reason-critic
A self-verification model that critiques agent output — it doesn't generate, it flags errors.
fableforge-14b
FableForge-14B: Four-stage training pipeline for a 14B coding agent model
error-recovery
Self-healing agent middleware that intercepts errors and injects recovery prompts
verifyloop
Agent framework implementing Plan → Execute → Verify → Recover with trained verification
