Back to skills
SKILL.md
Rlm Search
ASecurity3-Phase Knowledge Search strategy enforcing optimal lookup order - RLM Summary Scan -> Vector DB Semantic Search -> Grep/Exact Match.
- 7 stars
- 0 votes
- 0 copies
- 3 views
- Added September 2, 2026
Security analysis
100/100Pro scans all 20 files and shows the line behind each finding
npx -y skills add richfrem/agent-plugins-skills --skill rlm-search --agent claude-codeAre you the author of Rlm Search?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/richfrem-rlm-search-agent-plugins-skills)---
name: rlm-search
plugin: agent-memory
description: 3-Phase Knowledge Search strategy enforcing optimal lookup order - RLM Summary Scan -> Vector DB Semantic Search -> Grep/Exact Match.
allowed-tools: Bash, Read, Write
---
# RLM Knowledge Search (`rlm-search`)
Enforces the 3-phase knowledge search protocol to locate repository context with minimal token overhead.
## Contents
- [Critical Constraints](#critical-constraints)
- [Quick start](#quick-start)
- [Search Order Protocol](#search-order-protocol)
- [Workflow](#workflow)
- [Verification](#verification)
- [References](#references)
## Critical Constraints
1. **Start at Phase 1**: Never skip directly to full-repo grep or read files cold without checking summary cache first.
2. **Scope Grep Searches**: Restrict regex or grep searches to files identified in Phase 1 or Phase 2.
3. **Never Edit Ledgers**: Reading summaries is purely non-destructive.
## Quick start
Execute a Phase 1 scan across pre-computed summary caches:
```bash
python3 scripts/query_cache.py "session memory persistence" --profile wiki
```
## Search Order Protocol
```
Phase 1: RLM Summary Scan -> O(1), ~1ms -> Table of Contents
Phase 2: Vector DB Semantic -> O(log N), 1-5s -> Back-of-book Index
Phase 3: Scoped Exact Grep -> O(N), scoped -> Ctrl+F
```
## Workflow
1. **Phase 1 (RLM Scan)**: Scan summaries using `scripts/query_cache.py` or `grep_search .agent/learning/*_cache/`. Stop if answered.
2. **Phase 2 (Vector Search)**: If ranked similarity across chunks is needed, query `scripts/query.py "query" --profile wiki`. Stop if answered.
3. **Phase 3 (Scoped Grep)**: If exact symbol or function names are required, run targeted `rg "pattern" <scoped-dir>`.
4. **Transparent Citations**: State which search phase resolved the target and provide exact file links.
## Verification
Confirm query tool responsiveness and cache availability:
```bash
python3 scripts/query_cache.py "health" --profile wiki
```
## References
- [acceptance-criteria.md](references/acceptance-criteria.md) — Acceptance criteria for 3-phase search retrieval accuracy.
- [cheapest_models.md](references/cheapest_models.md) — Model tiers and pricing guidance for semantic query engines.
- [RLM_ARCHITECTURE.md](references/RLM_ARCHITECTURE.md) — Architectural overview of the 3-phase lookup design.
Files in this skill
- SKILL.md
- acceptance-criteria.md
- assets/diagrams/rlm-factory-architecture.mmd
- assets/diagrams/rlm-factory-dual-path.mmd
- assets/diagrams/rlm_late_binding_flow.mmd
- assets/diagrams/rlm_tool_enrichment_flow.mmd
- assets/diagrams/search_process.mmd
- assets/resources/distiller_manifest.json
- assets/resources/manifest-index.json
- assets/resources/prompts/rlm/rlm_summarize_general.md
- assets/resources/prompts/rlm/rlm_summarize_tool.md
- assets/resources/rlm_manifest.json
- evals/evals.json
- evals/results.tsv
- references/BLUEPRINT.md
- references/RLM_ARCHITECTURE.md
- references/acceptance-criteria.md
- references/cheapest_models.json
- references/cheapest_models.md
- references/diagrams/distillation_process.mmd
Attribution
Comments
Loading comments…