{
  "id": "page:docs-reference-repos-memodb-io-acontext-research",
  "_kind": "Page",
  "_file": "wiki/docs/reference-repos/memodb-io/acontext/research.md",
  "_cluster": "wiki",
  "attributes": {
    "nodeKind": "Page",
    "sourcePath": "docs/reference-repos/memodb-io/Acontext/research.md",
    "sourceKind": "repo-docs",
    "title": "memodb-io/Acontext",
    "displayName": "memodb-io/Acontext",
    "slug": "docs/reference-repos/memodb-io/acontext/research",
    "articlePath": "wiki/docs/reference-repos/memodb-io/Acontext/research.md",
    "article": "\n# memodb-io/Acontext\n\n- **Archetype**: context-platform\n- **Stars**: 3,314\n- **Last pushed**: 2026-04-11\n- **License**: Apache-2.0\n- **Discovered**: 2026-04-12\n- **Skills found**: 3\n\n## Summary\n\nA 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.\n\n## Assessment\n\nHigh 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.\n\n## Extraction Priority\n- Medium\n- 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.\n\n## Processes\n\n### 1. Self-Evolving Process Template\n- **Source skills**: daily-logs template, self-learning concept\n- **Placement**: `methodologies/self-evolving-process.js`\n- **Description**: Meta-process that wraps any process with a learning loop: execute -> log outcomes -> extract patterns -> update process heuristics for next run.\n\n### 2. Context Accumulation Workflow\n- **Source skills**: daily-logs, user-general-facts\n- **Placement**: `specializations/shared/context-accumulation.js`\n- **Description**: Structured context building process: daily activity logging -> fact extraction -> profile building -> context injection for subsequent runs.\n\n## Plugin Ideas\n\n- **Memory Layer plugin**: Persistent skill-scoped memory that accumulates context across runs (daily logs, learned facts, user preferences). Category: context & memory.\n- **Self-Evolving Skills plugin**: Track skill effectiveness across runs and auto-suggest skill refinements based on outcome patterns. Category: knowledge management.\n\n## Patterns\n\n- Skills as persistent memory (not just instructions, but accumulating knowledge)\n- Daily-log skill template: one file per day, append-only, ISO date naming\n- Third-person user references in logs (\"The user requested X\") for objectivity\n- Skill templates shipped with the platform as starting points\n\n## Library Mapping\n\n| Extractable Process | Library Status | Action | Existing Path | Target Placement |\n|-------------------|----------------|--------|---------------|------------------|\n| Self-Evolving Process Template | NEW | Meta-process that wraps any process with learning loop and outcome-based adaptation | - | methodologies/self-evolving-process/ |\n| Context Accumulation Workflow | NEW | Structured context building with daily logging and fact extraction | - | specializations/shared/context-accumulation-workflow.js |\n| Daily Activity Logging Process | NEW | Structured daily logging with append-only entries and ISO date naming | - | specializations/shared/daily-activity-logging.js |\n| User Profile Building | NEW | Fact extraction and user context profile generation from activity logs | - | specializations/shared/user-profile-building.js |\n| Skills-as-Memory Pattern | NEW | Persistent skill-scoped memory accumulation across multiple runs | - | specializations/shared/skills-as-memory-pattern.js |\n| Skill Template Management | NEW | Template system for bootstrapping new skills with learning capabilities | - | specializations/shared/skill-template-management.js |\n| Process Heuristic Adaptation | NEW | Extract patterns from process outcomes to improve future executions | - | specializations/shared/process-heuristic-adaptation.js |\n\n## Plugin Marketplace Mapping\n\n| Plugin Idea | Marketplace Status | Action | Existing Plugin | Target Placement |\n|-------------|-------------------|--------|-----------------|------------------|\n| Memory Layer | NEW | Persistent skill-scoped memory that accumulates context across runs | claude-mem | plugins/a5c/marketplace/blueprints/memory-layer/ |\n| Self-Evolving Skills Tracker | NEW | Track skill effectiveness and suggest refinements based on outcome patterns | - | plugins/a5c/marketplace/blueprints/self-evolving-skills-tracker/ |\n",
    "documents": []
  },
  "outgoingEdges": [],
  "incomingEdges": [
    {
      "from": "page:docs-reference-repos",
      "to": "page:docs-reference-repos-memodb-io-acontext-research",
      "kind": "contains_page"
    }
  ]
}