offerguard-mcp-server 0.3.1 → 0.4.2

package/README.md

@@ -1,102 +1,13 @@
-# OfferGuard MCP Server
+# OfferGuard MCP 0.4.2
 
-OfferGuard MCP Server provides real MCP tools for internship offer risk auditing, compliance checks, seven-dimension scoring, and workflow routing.
+This package bundles the real official MCP SDK and OfferGuard rules for reproducible startup. It does not simulate tools or delegate scoring to a language model. Node.js >=20 is required. No dependency download is required after installing this package.
 
-## Tools
+## Run
 
-- `offerguard_audit`: all-in-one workflow tool for XFYun Agent workflow nodes. Input: `raw_input`.
-- `risk_term_matcher`: matches risky JD terms and returns evidence.
-- `internship_compliance_checker`: checks internship red lines such as unpaid trials, deposits, training loans, work hours, salary payment, and IP ownership.
-- `offer_score_calculator`: calculates seven-dimension scores, overall score, risk level, and route.
+- stdio: `npx -y offerguard-mcp-server@0.4.2` after this version is published.
+- HTTP/SSE: `npm start` in the installed package directory. Default bind address: 127.0.0.1:3000.
+- Tools: offerguard_audit, risk_term_matcher, internship_compliance_checker, offer_score_calculator.
+- Input: raw_input; the original AGENT_USER_INPUT name remains supported. Both names accept labeled source text or the structured JSON envelope. Conflicting values are rejected.
+- Original source is in src; runtime files are in dist. Build with the matching repository's package-lock.json and scripts/build-hosted-package.js. Third-party licenses are in THIRD_PARTY_NOTICES.txt.
 
-## Recommended XFYun Workflow Usage
-
-For XFYun workflow MCP nodes, prefer the single entry tool:
-
-```json
-{
-  "tool": "offerguard_audit",
-  "arguments": {
-    "raw_input": "{{开始 / AGENT_USER_INPUT}}"
-  }
-}
-```
-
-The tool returns structured JSON with these key fields:
-
-- `mcp_ok`
-- `route`
-- `risk_level`
-- `overall_score`
-- `route_marker`
-- `route_real_marker`
-- `risk_terms`
-- `compliance_red_lines`
-- `compliance_missing_items`
-- `dimensions`
-- `required_followup_questions`
-- `negotiation_cards`
-- `accept_conditions`
-- `mcp_audit_summary`
-
-If the workflow brancher needs a text marker, read either:
-
-- `ROUTE: reject`
-- `ROUTE_REAL: compare`
-- `ROUTE_REAL: negotiate`
-
-## Local Run
-
-```bash
-npm install
-npm test
-npm start
-```
-
-Streamable HTTP endpoint:
-
-```text
-http://127.0.0.1:3000/mcp
-```
-
-Legacy SSE endpoints:
-
-```text
-http://127.0.0.1:3000/sse
-http://127.0.0.1:3000/messages
-```
-
-Health check:
-
-```text
-http://127.0.0.1:3000/health
-```
-
-## NPM Usage
-
-```bash
-npx -y offerguard-mcp-server@0.3.1
-```
-
-For MCP stdio clients:
-
-```json
-{
-  "mcpServers": {
-    "OfferGuard": {
-      "command": "npx",
-      "args": ["-y", "offerguard-mcp-server@0.3.1"]
-    }
-  }
-}
-```
-
-## Test Coverage
-
-`npm test` verifies:
-
-- Streamable HTTP initialization and `tools/list`.
-- stdio initialization and `tools/list`.
-- Calls to all four tools.
-- Low/medium/high-risk demo cases.
-- `offerguard_audit` uses workflow-safe `raw_input`, bound to `Start / AGENT_USER_INPUT` in XFYun.
+A local package build does not prove successful cloud hosting. No current deployment claim is made.