memodb-io/Acontext

• **Archetype**: context-platform
• **Stars**: 3,314
• **Last pushed**: 2026-04-11
• **License**: Apache-2.0
• **Discovered**: 2026-04-12
• **Skills found**: 3

Summary

A TypeScript/Go platform providing "Agent Skills as a Memory Layer." Acontext positions agent skills as persistent, self-evolving context that learns from usage. Includes a Go backend API, landing page, and skill template system. The 3 SKILL.md files are templates (daily-logs, user-general-facts) and a landing page skill. Topics emphasize context engineering, self-learning, agent observability, and LLMOps. The platform treats skills as living documents that accumulate knowledge over time.

Assessment

High conceptual value. The "skills as memory" paradigm aligns well with babysitter's event-sourced model -- skills that evolve based on run history. The daily-logs skill template demonstrates a structured journaling pattern (one file per day, append-only entries) that maps to babysitter's journal model. The user-general-facts template shows how to maintain persistent user context. The self-evolving skill concept could inform how babysitter processes adapt based on retrospective feedback.

Extraction Priority

• Medium
• Rationale: The platform architecture (Go backend, context data platform) is not directly extractable, but the skill-as-memory pattern and skill templates are conceptually valuable. The daily-logs and user-general-facts templates could inform babysitter context/memory plugins.

Processes

1. Self-Evolving Process Template

• **Source skills**: daily-logs template, self-learning concept
• **Placement**: methodologies/self-evolving-process.js
• **Description**: Meta-process that wraps any process with a learning loop: execute -> log outcomes -> extract patterns -> update process heuristics for next run.

2. Context Accumulation Workflow

• **Source skills**: daily-logs, user-general-facts
• **Placement**: specializations/shared/context-accumulation.js
• **Description**: Structured context building process: daily activity logging -> fact extraction -> profile building -> context injection for subsequent runs.

Plugin Ideas

• **Memory Layer plugin**: Persistent skill-scoped memory that accumulates context across runs (daily logs, learned facts, user preferences). Category: context & memory.
• **Self-Evolving Skills plugin**: Track skill effectiveness across runs and auto-suggest skill refinements based on outcome patterns. Category: knowledge management.

Patterns

• Skills as persistent memory (not just instructions, but accumulating knowledge)
• Daily-log skill template: one file per day, append-only, ISO date naming
• Third-person user references in logs ("The user requested X") for objectivity
• Skill templates shipped with the platform as starting points

Library Mapping

Plugin Marketplace Mapping