registry / AI-agent capability abuse / @respira/wordpress-mcp-server / 8.4.11

@respira/wordpress-mcp-server@8.4.11

WordPress MCP server + companion plugin giving AI native access to page builder content, not just the REST API. 248 tools, 352 with WooCommerce, 17 page builders and the Site Editor. Element-level edits, full page creation, HTML to builder conversion, sna

AI Security Review

scanned 12d ago · by lpm-firewall-ai

The package has a user-invoked agent configuration installer that writes its own server entry, so this is a guarded agent capability that warrants a warning. The inspected sources do not establish unconsented install-time mutation or concrete attack behavior.

Static reason

No blocking static signals were detected.; previous stored version diff introduced dangerous source

Trigger

A user runs setup with the install-config, client, and install-token arguments.

Impact

The agent client can start the Respira WordPress MCP server after configuration is installed.

Mechanism

The command fetches a client-specific configuration and merges the named Respira server entry into that client's configuration file.

AI rationale

Evidence

dist/setup.jsdist/install-skills.js

Decision evidence

public snapshot

AI called this Suspicious at 83.0% confidence as Dangerous Capability with medium false-positive risk.

Evidence for warning

The manifest exposes separate server, setup, and skill-installer commands.
The cited config.js location describes configuration precedence; runtime loading consumes explicit environment configuration.
The skill installer takes its destination from command-line options and copies bundled skills when invoked.
The setup command accepts a client and install token, then fetches configuration for that client.
The installer merges the named Respira server entry into the selected client configuration and writes the result.

Evidence against

Behavioral surface

Source

ChildProcessCryptoEnvironmentVarsFilesystemNetworkShell

Supply chain

HighEntropyStringsPossibleObfuscationUrlStrings

ManifestNo manifest risk signals triggered.

scanned 31 file(s), 1.36 MB of source, external domains: acme.test, example.com, nodejs.org, registry.npmjs.org, respira.press, www.respira.press, yoursite.com

Source & flagged code

6 flagged

dist/install-skills.jsView file

Lines 19-59javascript

19import * as path from 'path';

20import * as os from 'os';

21import { fileURLToPath } from 'url';

22const __filename = fileURLToPath(import.meta.url);

23const __dirname = path.dirname(__filename);

24// Resolve the bundled skills directory relative to the compiled JS.

25// dist/install-skills.js -> ../skills (sibling to dist/ inside the package)

26const BUNDLED_SKILLS_DIR = path.resolve(__dirname, '..', 'skills');

27const c = {

28 reset: '\x1b[0m',

29 bold: '\x1b[1m',

30 dim: '\x1b[2m',

31 red: '\x1b[31m',

32 green: '\x1b[32m',

33 yellow: '\x1b[33m',

34 blue: '\x1b[34m',

35 cyan: '\x1b[36m',

36};

37function parseArgs(argv) {

38 const opts = {

39 dest: path.join(os.homedir(), '.claude', 'skills'),

40 force: false,

41 help: false,

42 };

43 for (let i = 0; i < argv.length; i++) {

44 const arg = argv[i];

45 if (arg === '--help' || arg === '-h') {

Medium

Ai Review Evidence

The skill installer takes its destination from command-line options and copies bundled skills when invoked.

dist/install-skills.jsView on unpkg · L39

46 opts.help = true;

47 }

48 else if (arg === '--force' || arg === '-f') {

49 opts.force = true;

50 }

51 else if (arg === '--dest' || arg === '-d') {

52 const next = argv[i + 1];

53 if (!next || next.startsWith('-')) {

54 console.error(`${c.red}error:${c.reset} --dest requires a path argument`);

55 process.exit(2);

56 }

57 opts.dest = path.resolve(next);

58 i++;

59 }

•Manifest-trigger-reachable source links an external AI-agent control path to a behavior-bearing write operation. const bundled_skills_dir = path.resolve(__dirname, '..', 'skills'); const c = { reset: '\x1b[0m', bold: '\x1b[1m', dim: '\x1b[2m', red: '\x1b[31m', green: '\x1b[32m', yellow: '\x1b[33m', blue: '\x1b[34m', cyan: '\x1b[36m', }; function parseargs(argv) { const opts = { dest: path.join(os.homedir(), '.claude', 'skills'), force: false, help: false, }; for (let i = 0; i < argv.length; i++) { const arg = argv[i]; if (arg === '--help' || arg === '-h') { opts.help = true; } else if (arg === '--force' || arg === '-f') { opts.force = true; } else if (

High

Trigger Reachable External Ai Agent Control Surface Mutation

Manifest-trigger-reachable source writes behavior-bearing configuration into a user or project AI-agent control surface.

dist/install-skills.jsView on unpkg

dist/config.js#virtual:normalized:round1View file

•stage = ast_semantic_analysis; reason = ast_parse_error; limitedFiles = 2

High

Semantic Analysis Limited

A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.

dist/config.js#virtual:normalized:round1View on unpkg

dist/server.jsView file

•matchType = previous_version_dangerous_delta matchedPackage = @respira/wordpress-mcp-server@8.3.24 matchedIdentity = npm:[redacted]:8.3.24 similarity = 0.875 summary = stored previous version shares package body but lacks this dangerous source file

High

Previous Version Dangerous Delta

This package version adds a dangerous source file absent from the previous stored version; route for source-aware review.

dist/server.jsView on unpkg

package.jsonView file

Lines 1-26json

2 "name": "@respira/wordpress-mcp-server",

3 "version": "8.4.11",

4 "mcpName": "io.github.webmyc/respira-wordpress",

5 "description": "WordPress MCP server + companion plugin giving AI native access to page builder content, not just the REST API. 248 tools, 352 with WooCommerce, 17 page builders and the Site Editor. Element-level edits, full page creation, HTML to builder conversion, snapshots and rollback on every write. MCP 2026-07-28 with legacy-client support. New in 8.3: Design Direction, one design system per site, applied to native builder stores, checked against the rendered page, with a live preview link for the Art Direction dashboard."

dist/config.jsView file

Lines 756-796javascript

756 * Resolution order is RESPIRA_CONFIG_B64, then RESPIRA_CONFIG_FILE, then

757 * ~/.respira/config.json. Only the winner matters, and until now nothing

758 * reported which one won. That gap is the common factor behind three separate

759 * customer incidents whose mechanisms were all different: a redeem that wrote a

Source & flagged code

6 flagged · loading source

dist/install-skills.jsView file

High

Trigger Reachable External Ai Agent Control Surface Mutation

Manifest-trigger-reachable source writes behavior-bearing configuration into a user or project AI-agent control surface.

dist/install-skills.jsView on unpkg

39'.claude', 'skills'), force: false, help: false, }; for (let i = 0; i < argv.length; i++) { const arg = argv[i]; if (arg === '--help' || arg

Medium

Ai Review Evidence

The skill installer takes its destination from command-line options and copies bundled skills when invoked.

dist/install-skills.jsView on unpkg · L39

Findings

3 High5 Medium4 Low

HighTrigger Reachable External Ai Agent Control Surface Mutationdist/install-skills.js

HighSemantic Analysis Limiteddist/config.js#virtual:normalized:round1

HighPrevious Version Dangerous Deltadist/server.js

MediumNetwork

MediumEnvironment Vars

MediumAi Review Evidencepackage.json

MediumAi Review Evidencedist/config.js

MediumAi Review Evidencedist/install-skills.js

LowScripts Present

LowFilesystem

LowHigh Entropy Strings

LowUrl Strings

Affected versions and remediation

This report applies to @respira/wordpress-mcp-server@8.4.11.

See all versions of @respira/wordpress-mcp-server for other recorded verdicts.

Review the evidence and your use of @respira/wordpress-mcp-server@8.4.11 before allowing it. Restrict the permissions described in this report.
Choose an independently verified alternative or release. This report does not establish that other versions are safe.

Evidence last updated: 2026-10-11 04:20:10.717Z (UTC).

Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Runtime surface

Behavioral surface

Source & flagged code