skills/zz3310969/max-skills/mcp-us-equities-ops

mcp-us-equities-ops

SKILL.md

MCP US Equities Ops

Purpose

Use this skill for operational tools on the VPS MCP service:

  • list_pipelines
  • run_pipeline
  • track_run
  • health_check

Prerequisites

  1. Install and configure rw to the VPS endpoint:
rw setup --server-url http://113.44.56.214:18080/mcp/
rw doctor
  1. Use direct calls only:
rw call --tool <tool_name> --args '<json>'

Preflight (required before any rw call)

Run this guard first to avoid stale rw versions:

if ! command -v rw >/dev/null 2>&1; then
  curl -fsSL https://raw.githubusercontent.com/zz3310969/max-skills/main/scripts/install-rw.sh | bash
fi
rw --version || true
rw doctor

If rw doctor fails due to protocol/auth/session issues, force reinstall:

curl -fsSL https://raw.githubusercontent.com/zz3310969/max-skills/main/scripts/install-rw.sh | bash
rw doctor

Operational Workflow

  1. Discover deployment names using list_pipelines.
  2. Validate trigger payload with run_pipeline and dry_run=true.
  3. Execute with optional idempotency_key.
  4. Poll status with track_run.
  5. Pull summarized logs only when needed.

Commands

List pipelines

rw call --tool list_pipelines --args '{"name_like":"barchart","limit":100}'

Dry run pipeline

rw call --tool run_pipeline --args '{"deployment_name":"Market API Daily Technicals/daily-barchart-technicals","params":{"ticker":"NVDA"},"dry_run":true,"idempotency_key":"dryrun-20260309-nvda"}'

Execute pipeline (idempotent)

rw call --tool run_pipeline --args '{"deployment_name":"Market API Daily Technicals/daily-barchart-technicals","params":{"ticker":"NVDA"},"dry_run":false,"idempotency_key":"run-20260309-nvda"}'

Track run status

rw call --tool track_run --args '{"run_id":"<FLOW_RUN_ID>","include_logs":false}'

Track run with compressed logs

rw call --tool track_run --args '{"run_id":"<FLOW_RUN_ID>","include_logs":true,"log_lines":120}'

Health checks

rw call --tool health_check --args '{"checks":["db","prefect_server","prefect_worker","market_api"]}'

Safety Rules

  • Always start with dry_run=true for manual triggers.
  • Use stable idempotency_key to avoid duplicate runs.
  • If error.code=INVALID_MARKET_SCOPE in run_pipeline, remove non-US tickers from params and retry.
  • If error.retryable=true, use bounded retry/backoff.

Reference mapping: docs/mcp-tool-map-v2.md.

Weekly Installs
4
First Seen
7 days ago
Installed on
openclaw4
gemini-cli4
claude-code4
github-copilot4
codex4
kimi-cli4