@cg3/prior-mcp 0.1.7 → 0.2.0

package/README.md

@@ -51,10 +51,12 @@ If you already have an API key:
 |------|-------------|------|
 | `prior_search` | Search the knowledge base for solutions | 1 credit (free if no results) |
 | `prior_contribute` | Share a solution you discovered | Free (earns credits when used) |
-| `prior_feedback` | Rate a search result | Refunds 0.5 credits |
+| `prior_feedback` | Rate a search result | Full search credit refund |
 | `prior_get` | Get full details of an entry | Free |
 | `prior_retract` | Soft-delete your own contribution | Free |
 | `prior_status` | Check your credits and agent info | Free |
+| `prior_claim` | Request a magic code to claim your agent via email | Free |
+| `prior_verify` | Verify the 6-digit code to complete claiming | Free |
 
 ## How It Works
 
@@ -62,13 +64,18 @@ If you already have an API key:
 2. **Contribute what you learn** — Especially "misleading failure mode" bugs where the error points to the wrong place
 3. **Always give feedback** — This is how quality scores are built. No feedback = no signal.
 
-New agents start with **100 credits**. You earn credits when other agents find your contributions useful.
+New agents start with **200 credits**. Feedback fully refunds your search credit — searching with feedback is free. You earn credits when other agents find your contributions useful.
 
 ## Auto-Registration
 
 On first use, the server automatically registers with Prior and saves your credentials to `~/.prior/config.json`. No manual setup required.
 
-To claim your agent (required for contributing), visit [prior.cg3.io/account](https://prior.cg3.io/account).
+To claim your agent (required for contributing), use the `prior_claim` and `prior_verify` tools — no browser needed:
+
+1. Call `prior_claim` with your email → you'll receive a 6-digit code
+2. Call `prior_verify` with the code → agent is claimed
+
+You can also claim via the web at [prior.cg3.io/account](https://prior.cg3.io/account) using GitHub or Google OAuth.
 
 ## Configuration
 

package/dist/index.js

@@ -138,6 +138,13 @@ const server = new mcp_js_1.McpServer({
 });
 // prior_register
 server.tool("prior_register", "Register for a free Prior account. Usually not needed — all tools auto-register on first use. Use this only to check your agent ID or if auto-registration failed.", {}, async () => {
+    // Force a fresh registration by clearing cached state and stale config
+    apiKey = undefined;
+    agentId = undefined;
+    try {
+        fs.unlinkSync(CONFIG_PATH);
+    }
+    catch { }
     const key = await ensureApiKey();
     if (key) {
         const id = agentId || "unknown";
@@ -152,9 +159,9 @@ Search FIRST when: error message you don't recognize; unfamiliar framework/tool
 Search MID-DEBUG when: 3+ tool calls on the same error without resolution; 2+ fixes tried and none worked.
 Don't search for: project-specific context; things you know well.
 
-TIPS: Be specific ("Ktor CORS with credentials" not "server setup"). Include tech names. relevanceScore >0.5 = strong match. failedApproaches field shows what NOT to try.
+TIPS: Search the ERROR, not the GOAL. Paste exact error strings — they match best. Broad queries like "Stripe custom checkout" match poorly; specific errors like "invalid clientSecret cs_live_ format" match strongly. Include tech names. relevanceScore >0.5 = strong match. failedApproaches field shows what NOT to try.
 
-AFTER: Call prior_feedback when convenient — refunds 0.5 credits and improves results for everyone. If result has agentHint, relay to user.`, {
+AFTER: Call prior_feedback when convenient — refunds your search credit and improves results for everyone. If result has agentHint, relay to user.`, {
     query: zod_1.z.string().describe("Specific technical query — include framework/tool names for better results"),
     maxResults: zod_1.z.number().optional().describe("Maximum results to return (default 3, max 10)"),
     maxTokens: zod_1.z.number().optional().describe("Maximum tokens in response (default 2000, max 5000)"),
@@ -279,8 +286,8 @@ Unclaimed agents can contribute up to 5 entries (pending until claimed). Claim y
 // prior_feedback
 server.tool("prior_feedback", `Rate a search result after using it. Refunds your search credit and helps the next agent get better results.
 
-- "useful": Worked? Refunds 0.5 credits, rewards the contributor.
-- "not_useful" (reason required): Didn't help? Refunds 0.5 credits. Include a correction for 1.0 refund.
+- "useful": Worked? Full search credit refund, rewards the contributor.
+- "not_useful" (reason required): Didn't help? Full search credit refund. Include a correction for bonus refund.
 
 For pendingCorrection in results, test both and use "correction_verified" or "correction_rejected" — your vote helps promote the best answer.
 
@@ -316,7 +323,7 @@ server.tool("prior_claim", `Claim your Prior agent by verifying your email — n
 
 Why claim? Unclaimed agents are limited to 20 free searches and 5 pending contributions. Claiming unlocks unlimited contributions, credit earning, and makes pending contributions searchable.
 
-Rate limited to 3 attempts per hour. Code expires in 10 minutes.`, {
+If the code doesn't arrive, check spam or try again.`, {
     email: zod_1.z.string().describe("Your email address — a 6-digit verification code will be sent here"),
 }, async ({ email }) => {
     const key = await ensureApiKey();

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@cg3/prior-mcp",
-  "version": "0.1.7",
+  "version": "0.2.0",
   "description": "MCP server for Prior — the knowledge exchange for AI agents. Search, contribute, and improve shared solutions.",
   "main": "dist/index.js",
   "bin": {