{
  "id": "page:docs-reference-repos-clawhub-fajarhide-omni-research",
  "_kind": "Page",
  "_file": "wiki/docs/reference-repos/clawhub/fajarhide/omni/research.md",
  "_cluster": "wiki",
  "attributes": {
    "nodeKind": "Page",
    "sourcePath": "docs/reference-repos/clawhub/fajarhide/omni/research.md",
    "sourceKind": "repo-docs",
    "title": "fajarhide/omni",
    "displayName": "fajarhide/omni",
    "slug": "docs/reference-repos/clawhub/fajarhide/omni/research",
    "articlePath": "wiki/docs/reference-repos/clawhub/fajarhide/omni/research.md",
    "article": "\n# fajarhide/omni\n\n- **Archetype**: clawhub-plugin\n- **Stars**: 49\n- **Last pushed**: 2026-04-12\n- **License**: MIT\n- **Discovered**: 2026-04-12\n- **Language**: Rust\n- **Fork**: No\n\n## Summary\n\nOMNI is a Rust-based terminal output filter that intercepts command output before it reaches the AI agent, removing noise and preserving signal. Claims up to 90% token reduction. Integrates via native hooks (pre-hook rewriter, post-hook distiller, session-start context, pre-compact summary) into Claude Code, OpenClaw, and MCP agents.\n\nCore architecture: Registry (TOML filter definitions per tool) -> Scorer (context boost) -> Distiller (semantic filtering). Persistence via SQLite for session state and a \"RewindStore\" that archives raw output for on-demand retrieval. Ships with 35+ tool-specific filter definitions (vitest, jest, eslint, tsc, cargo, docker, kubectl, terraform, etc.) in TOML format.\n\nKey features: DSL-based filter configuration (capture/action/output templates), session intelligence (tracks active files to avoid redundant context), RewindStore for lossless raw output access, `omni stats` / `omni diff` for token savings visualization.\n\n## Assessment\n\nHIGH extractable value. OMNI's approach to token compression via tool-specific output filtering is directly relevant to babysitter's existing compression layer (`packages/sdk/src/compression/`). The TOML filter registry pattern, capture/action DSL, and RewindStore concept are all adaptable. The 35+ pre-built filter definitions for common dev tools are immediately useful.\n\n**Extraction priority**: HIGH\n\n# Extractable Value: fajarhide/omni\n\n## Processes\n\nNone -- OMNI is a tool, not a workflow. Its value is entirely in plugin patterns.\n\n## Plugin Ideas\n\n### 1. Smart Output Filter Plugin\n- **Category**: Context & Memory\n- **install.md**: Installs a hook-based output filter that intercepts tool execution results before they enter the agent context. Uses a TOML-based filter registry with per-tool definitions (capture patterns, action rules, output templates). Includes a RewindStore that archives raw output for on-demand retrieval when the agent needs full detail. Ships with pre-built filters for 35+ common dev tools (vitest, jest, eslint, tsc, cargo, docker, kubectl, terraform, npm, pip, etc.).\n\n### 2. Token Savings Dashboard Plugin\n- **Category**: Developer Experience & UX\n- **install.md**: Adds `babysitter compression:stats` enhancement showing per-tool token savings over time. Tracks raw vs filtered output sizes, calculates cost savings, provides `diff` visualization comparing raw output to filtered version. Persists metrics in SQLite for historical analysis.\n\n## Library Mapping\n\n| Extractable Process | Library Status | Action | Existing Path | Target Placement |\n|-------------------|----------------|--------|---------------|------------------|\n| N/A | N/A | No multi-step processes identified - OMNI is a tool, not a workflow | - | N/A |\n\n## Plugin Marketplace Mapping\n\n| Plugin Idea | Marketplace Status | Action | Existing Plugin | Target Placement |\n|-------------|-------------------|--------|-----------------|------------------|\n| Smart Output Filter | NEW | Hook-based output filter with TOML registry and RewindStore for lossless compression | - | plugins/a5c/marketplace/plugins/smart-output-filter/ |\n| Token Savings Dashboard | NEW | Compression stats enhancement with per-tool metrics and cost visualization | - | plugins/a5c/marketplace/plugins/token-savings-dashboard/ |\n\n## Implicit Procedural Knowledge\n\n- **TOML filter registry pattern**: Each dev tool gets its own TOML file defining capture patterns, confidence scores, and output templates. This is a clean, extensible pattern for babysitter's compression layer -- currently babysitter uses a density-filter engine with FNV-1a dedup, but OMNI's approach of tool-specific semantic filtering could complement it.\n- **DSL capture/action/output model**: The `omni_config.json` shows a simple but powerful DSL: `trigger` (fast string match), `capture` (template extraction with named fields), `action` (keep/count), `output` (template with extracted fields). This could inform a babysitter plugin API for custom compression rules.\n- **RewindStore pattern**: Archive raw output, serve filtered output by default, allow on-demand raw access. Prevents information loss while still saving tokens. Directly relevant to babysitter's `compress-output` command.\n- **Session intelligence**: Tracking which files are actively being edited to avoid feeding the agent context it already knows. This is a form of deduplication that goes beyond babysitter's current FNV-1a approach.\n",
    "documents": []
  },
  "outgoingEdges": [],
  "incomingEdges": [
    {
      "from": "page:docs-reference-repos",
      "to": "page:docs-reference-repos-clawhub-fajarhide-omni-research",
      "kind": "contains_page"
    }
  ]
}