@pseolint/mcp 0.6.2 → 0.6.3

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (2) hide show
  1. package/README.md +38 -4
  2. package/package.json +2 -2
package/README.md CHANGED
@@ -1,8 +1,38 @@
1
1
  # @pseolint/mcp
2
2
 
3
- > MCP server for pseolint — audit programmatic SEO sites from AI coding assistants.
3
+ > MCP server for pseolint — audit pSEO sites by template from AI coding assistants.
4
4
 
5
- An [MCP (Model Context Protocol)](https://modelcontextprotocol.io) server that exposes [pseolint](https://www.npmjs.com/package/pseolint) auditing tools to AI coding assistants like Claude Code, Claude Desktop, Cursor, and Windsurf.
5
+ An [MCP (Model Context Protocol)](https://modelcontextprotocol.io) server that exposes [pseolint](https://www.npmjs.com/package/pseolint) v0.6.2 auditing tools to AI coding assistants like Claude Code, Claude Desktop, Cursor, and Windsurf.
6
+
7
+ ### What's new in v0.6 — per-template output in tool responses
8
+
9
+ `audit_site` now returns a `templates` array alongside the existing `findings` list. AI clients get the per-template view automatically — no parameter changes needed.
10
+
11
+ ```json
12
+ {
13
+ "verdict": "concerning",
14
+ "risk": 60,
15
+ "templates": [
16
+ {
17
+ "signature": "/listing/:slug",
18
+ "totalUrls": 8201,
19
+ "auditedUrls": ["https://example.com/listing/foo", "..."],
20
+ "verdict": "concerning",
21
+ "risk": 60,
22
+ "variance": {
23
+ "uniformityScore": 0.85,
24
+ "topDriver": { "ruleId": "spam/thin-content", "fireRate": 0.8 }
25
+ }
26
+ },
27
+ { "signature": "/category/:slug", "verdict": "ready", "risk": 12 }
28
+ ],
29
+ "findings": [...]
30
+ }
31
+ ```
32
+
33
+ When iterating `templates`, the `topDriver` rule + `fireRate` is the most actionable signal: "8/10 samples fail `spam/thin-content`" tells the LLM which template is broken and what to fix. The `findings` flat list remains available for per-URL drill-down.
34
+
35
+ Design rationale: [`docs/superpowers/specs/2026-05-04-pseolint-v0.6-audit-as-template-reframe.md`](../../docs/superpowers/specs/2026-05-04-pseolint-v0.6-audit-as-template-reframe.md)
6
36
 
7
37
  ### What's new in v0.5.2 — credibility layer
8
38
 
@@ -52,15 +82,19 @@ Drive an LLM through 25 audit tools and produce a fix manifest with concrete pat
52
82
 
53
83
  ### `audit_site`
54
84
 
55
- Run a full pseolint audit on a URL or directory path. Returns the SpamBrain Risk Score (0-100) and all findings with actionable fix suggestions.
85
+ Run a full pseolint audit on a URL or directory path. Returns the site-level verdict + risk score, a `templates` array (per-template verdicts + variance metrics), and all per-URL findings with actionable fix suggestions.
56
86
 
57
87
  **Parameters:**
58
88
  - `source` (required) — URL or directory path to audit
59
89
  - `threshold` — Score threshold for pass/fail (default: 40)
60
90
  - `sampleSize` — Audit a random subset of N pages (0 = all)
61
91
  - `format` — Output format: `console` or `json` (default: console)
92
+ - `authorityScore` — 0-100 domain authority hint. ≥80 shifts verdict one tier lenient; ≤30 shifts one tier stricter.
93
+ - `sampleSeed` — Integer seed for deterministic stratified sampling. Same seed = same audit = same verdict.
94
+
95
+ **Returns (v0.6 shape):** `verdict`, `risk`, `categories`, `templates: Template[]`, `findings: RuleResult[]`. AI clients should iterate `templates` first — `topDriver.ruleId + fireRate` is the most actionable per-template signal. Use `findings` for per-URL drill-down.
62
96
 
63
- **Example prompt:** "Audit my site at http://localhost:3000 for SpamBrain risk"
97
+ **Example prompt:** "Audit my site at http://localhost:3000 show me the per-template breakdown"
64
98
 
65
99
  ### `explain_score`
66
100
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@pseolint/mcp",
3
- "version": "0.6.2",
3
+ "version": "0.6.3",
4
4
  "description": "MCP server for pseolint — audit programmatic SEO sites from AI coding assistants. v0.5 adds the AI-orchestrated audit tool that produces a fix manifest with concrete patches.",
5
5
  "license": "MIT",
6
6
  "author": "Ouranos Labs <contact@ouranos-labs.dev>",
@@ -43,7 +43,7 @@
43
43
  },
44
44
  "dependencies": {
45
45
  "@modelcontextprotocol/sdk": "^1.29.0",
46
- "@pseolint/core": "^0.6.2",
46
+ "@pseolint/core": "^0.6.3",
47
47
  "zod": "^4.3.6"
48
48
  }
49
49
  }