AgentStack
SKILL verified Apache-2.0 Self-run

Cbeta Research

skill-yellowbee686-everything-analysis-agent-cbeta-research · by yellowbee686

Use the CBETA MCP server for Chinese Buddhist scripture search, catalog lookup, metadata retrieval, KWIC inspection, and source-text citation. Use when Codex needs live CBETA queries, Buddhist canon text lookup, work/juan/linehead citation, translator or dynasty searches, or CBETA MCP tool guidance.

No reviews yet
0 installs
11 views
0.0% view→install

Install

$ agentstack add skill-yellowbee686-everything-analysis-agent-cbeta-research

✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

Are you the author of Cbeta Research? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

CBETA Research

Overview

Use the mcp_servers/CbetaMCP submodule as the live tool server. This skill provides startup policy, tool selection, and citation workflows; it does not duplicate the MCP tool implementations.

Startup Policy

Do not start the MCP server when the skill is loaded. Start it only when the user task needs a live CBETA query.

  1. Check the endpoint first:
python3 skills/cbeta-research/scripts/check_server.py --url http://localhost:18765/mcp/
  1. If unavailable, start on demand:
skills/cbeta-research/scripts/start_server.sh

When running inside an agent shell that cleans up background processes, keep the server in a long-running foreground session:

CBETA_MCP_FOREGROUND=1 skills/cbeta-research/scripts/start_server.sh
  1. Re-run the check. If it still fails, report dependency, port, or server-log diagnostics.

The default MCP endpoint is http://localhost:18765/mcp/.

Windows Preference

On Windows, prefer running CBETA MCP commands inside WSL2. Fall back to PowerShell only when WSL2 is unavailable or the user explicitly asks for native Windows execution.

  • If the agent is already running in WSL2, use the normal Linux commands above.
  • If the agent is running in native Windows PowerShell or CMD and WSL2 is available, run Linux commands through wsl.exe -- bash -lc.
  • If WSL2 is unavailable, use scripts\start_server.ps1 from PowerShell.
  • Convert Windows paths such as C:\Users\name\repo to WSL paths such as /mnt/c/Users/name/repo before calling Linux scripts.

Example from native Windows:

wsl.exe -- bash -lc 'cd /mnt/c/path/to/everything-analysis-agent && skills/cbeta-research/scripts/start_server.sh'

Fallback from native PowerShell:

.\skills\cbeta-research\scripts\start_server.ps1

Query Workflow

  1. Use references/tool-map.md to choose the smallest useful tool set.
  2. For broad discovery, start with cbeta_all_in_one or cbeta_search_sc.
  3. For exact Buddhist canon metadata, use catalog or work tools before content retrieval.
  4. For final answers, prefer source-backed snippets from get_cbeta_lines, cbeta_kwic_search, or get_juan_html.
  5. Include work, title, juan, and linehead when available. Preserve traditional Chinese source text unless the user asks for simplified Chinese.

References

  • Read references/tool-map.md for tool categories, common parameter shapes, and recommended flows.
  • Read references/setup.md for submodule initialization, dependency installation, endpoint configuration, and troubleshooting.

Output Rules

  • Distinguish search hits from verified source text.
  • Show CBETA identifiers in citations: work, title, juan, linehead, and URL from cbeta_goto when available.
  • Report MCP startup failures with the check/start command and relevant log path.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet — be the first.

Versions

  • v0.1.0 Imported from the upstream source.