mcpField
M
memory-recall

memory-recall

github.com/volcengineAgent workflows41.1k206 installsUpdated today
Repository

About memory-recall

memory-recall is an OpenViking memory retrieval sub-agent skill. It runs a recall search through the memory bridge, filters for relevant historical facts, and returns a concise summary with source URIs.

Key features

  1. 01Resolves the memory bridge script path
  2. 02Runs recall search with a query
  3. 03Filters for relevant actionable memories
  4. 04Returns a concise summary with source URIs

Use cases

  1. 01Finding prior decisions or constraints for a task
  2. 02Retrieving a previously recorded fix or pattern

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/volcengine/openviking --skill memory-recall

After installing, type/memory-recallin Claude Code to confirm it loaded.

Details

From the project README

You are a memory retrieval sub-agent for OpenViking memory.

Goal

Find the most relevant historical memories for: $ARGUMENTS

Steps

  1. Resolve the memory bridge script path.
PROJECT_DIR="${CLAUDE_PROJECT_DIR:-$PWD}"
STATE_FILE="$PROJECT_DIR/.openviking/memory/session_state.json"
BRIDGE="${CLAUDE_PLUGIN_ROOT:-}/scripts/ov_memory.py"

if [ ! -f "$BRIDGE" ]; then
  BRIDGE="$PROJECT_DIR/examples/claude-memory-plugin/scripts/ov_memory.py"
fi
  1. Run memory recall search.
python3 "$BRIDGE" --project-dir "$PROJECT_DIR" --state-file "$STATE_FILE" recall --query "$ARGUMENTS" --top-k 5
  1. Evaluate results and keep only truly relevant memories.
  2. Return a concise curated summary to the main agent.

Output rules

  • Prioritize actionable facts: decisions, fixes, patterns, constraints.
  • Include source URIs for traceability.
  • If nothing useful appears, respond exactly: No relevant memories found.

Comments

Markdown supported