Acontext and skill-conductor
These two tools are **complements**, where "Acontext" provides a memory layer for agent skills and "skill-conductor" offers a lifecycle management system to build, evaluate, and package those very skills, potentially leveraging "Acontext" to store and retrieve skill-related memory during its lifecycle.
About Acontext
memodb-io/Acontext
Agent Skills as a Memory Layer
Implements automatic skill extraction and progressive-disclosure retrieval without embeddings—using a "Skill Agent" to distill conversation traces into editable Markdown files that agents can fetch via tool calls. Supports Python and TypeScript SDKs, integrates with LangGraph and Claude, and persists skills as version-controllable files with optional self-hosting via Docker.
About skill-conductor
smixs/skill-conductor
Architecture-first skill lifecycle for AI agents. 5 modes: CREATE → EVAL → EDIT → REVIEW → PACKAGE. Integrates Anthropic's eval engine (grader/comparator/analyzer agents, blind A/B, benchmarks) with architecture patterns, TDD baseline, and 5-axis scoring. Not just testing - full design-to-distribution.
Bundles Anthropic's three-stage eval agents (grader, comparator, analyzer) with enforced architecture selection upfront—choosing from 5 patterns (Sequential, Iterative, Context-Aware, Domain Intelligence, Multi-MCP) before writing code. Runs TDD baselines in parallel to detect whether the task already works without a skill, then applies 5-axis scoring (Discovery/Clarity/Efficiency/Robustness/Completeness) with numerical thresholds for production readiness, plus automated description optimization with train/test splits.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work