@agent-lint/mcp 0.2.1 → 0.3.3

package/CHANGELOG.md

@@ -0,0 +1,38 @@
+# @agent-lint/mcp Changelog
+
+## [0.3.3]
+
+### Fixed
+
+- GitLab publish jobs now support both plain and file-based `NPM_TOKEN` variables and fail fast on invalid npm auth
+
+## [0.3.2]
+
+### Changed
+
+- README and npm description now use the current public positioning around structured, current, codebase-aware agent context
+- Public repository cleanup removed internal planning docs from the published surface
+
+## [0.3.1]
+
+### Changed
+
+- README, npm metadata, and release metadata now match the current 4-tool MCP surface
+
+### Fixed
+
+- Tool timeout wiring is now attached to the active public tools
+- Server version resolution now prefers the package version instead of unrelated workspace env values
+
+## [0.3.0]
+
+### Added
+
+- Read-only MCP server with 4 tools and 3 resources
+- Stdio and HTTP transports
+- Programmatic server creation helpers
+
+### Fixed
+
+- Current tool timeout wiring now matches the public tool surface
+- Server version reporting now resolves to the real package version

package/README.md

@@ -1,109 +1,39 @@
 # @agent-lint/mcp
 
-MCP server for AI agent context artifact analysis and quality scoring.
+Read-only MCP server for keeping `AGENTS.md`, rules, and skills structured, current, and codebase-aware.
 
-**Zero LLM dependencies.** Fully deterministic. Local-first. No database. No auth.
-
-## What It Does
-
-Provides a [Model Context Protocol](https://modelcontextprotocol.io/) server that evaluates AI-agent context artifacts:
-
-- `AGENTS.md` / `CLAUDE.md`
-- Skills, Rules, Workflows, Plans
-
-Reproducible quality scoring across **12 metrics**, evidence-backed assessment, guardrail checks, and repeatable improvement loops — all without calling any LLM.
-
-## Quick Start
+## Start the Server
 
 ```bash
 npx -y @agent-lint/mcp
 ```
 
-## MCP Client Configuration
-
-### Cursor
-
-Add to `.cursor/mcp.json`:
-
-```json
-{
-  "mcpServers": {
-    "agent-lint": {
-      "command": "npx",
-      "args": ["-y", "@agent-lint/mcp"]
-    }
-  }
-}
-```
-
-### Claude Desktop
+For most users, the fastest setup path is still:
 
-Add to `claude_desktop_config.json`:
-
-```json
-{
-  "mcpServers": {
-    "agent-lint": {
-      "command": "npx",
-      "args": ["-y", "@agent-lint/mcp"]
-    }
-  }
-}
-```
-
-### VS Code
-
-Add to your VS Code `settings.json`:
-
-```json
-{
-  "mcp": {
-    "servers": {
-      "agent-lint": {
-        "command": "npx",
-        "args": ["-y", "@agent-lint/mcp"]
-      }
-    }
-  }
-}
-```
-
-### Windsurf
-
-Add to your Windsurf MCP config:
-
-```json
-{
-  "mcpServers": {
-    "agent-lint": {
-      "command": "npx",
-      "args": ["-y", "@agent-lint/mcp"]
-    }
-  }
-}
+```bash
+npx @agent-lint/cli init
 ```
 
-## Available Tools
+## Current Tool Surface
 
-| Tool                           | Description                                                                          |
-| ------------------------------ | ------------------------------------------------------------------------------------ |
-| `analyze_artifact`             | Analyze a single artifact — returns scores across 12 metrics with findings and hints |
-| `analyze_workspace_artifacts`  | Scan a workspace directory for AI agent artifacts with framework detection           |
-| `analyze_context_bundle`       | Analyze multiple artifacts together for cross-artifact consistency                   |
-| `prepare_artifact_fix_context` | Prepare context for an artifact improvement loop                                     |
-| `submit_client_assessment`     | Submit a client LLM assessment with evidence-backed scores                           |
-| `quality_gate_artifact`        | Check if an artifact meets a target quality score                                    |
-| `suggest_patch`                | Generate patch suggestions to improve an artifact                                    |
-| `apply_patches`                | Apply patches to local files with hash guard, allowlist, and backup protection       |
-| `validate_export`              | Validate final artifact output for safety and correctness                            |
+| Tool | Purpose |
+| --- | --- |
+| `agentlint_get_guidelines` | Return artifact guidance before creating or updating context files |
+| `agentlint_plan_workspace_autofix` | Scan a workspace and return a step-by-step fix plan |
+| `agentlint_quick_check` | Decide whether recent code changes require context updates |
+| `agentlint_emit_maintenance_snippet` | Return a reusable maintenance rule snippet for supported IDEs |
 
-## Quality Metrics (12)
+## Current Resource Surface
 
-`clarity` · `specificity` · `scope-control` · `completeness` · `actionability` · `verifiability` · `safety` · `injection-resistance` · `secret-hygiene` · `token-efficiency` · `platform-fit` · `maintainability`
+| Resource | Purpose |
+| --- | --- |
+| `agentlint://guidelines/{type}` | Readable guidelines for one artifact type |
+| `agentlint://template/{type}` | Skeleton template for a new artifact |
+| `agentlint://path-hints/{type}` | IDE-specific file discovery hints |
 
 ## Programmatic Usage
 
-```typescript
+```ts
 import { createAgentLintMcpServer } from "@agent-lint/mcp";
 
 const server = createAgentLintMcpServer({
@@ -117,17 +47,23 @@ const server = createAgentLintMcpServer({
 npx @agent-lint/mcp --http --port 3001
 ```
 
-## Design Principles
+Notes:
+
+- HTTP mode keeps the public surface read-only.
+- Workspace scanning is disabled in HTTP mode unless you explicitly enable it.
+
+## Design Constraints
 
-- **No LLM**: Fully deterministic. No API calls, no tokens, no cost.
-- **No State**: Every call is stateless. No database, no cache.
-- **No File Writes**: Read-only by default. `apply_patches` requires explicit flags + hash guard.
-- **Minimum Dependencies**: Published package under 500 KB packed.
+- No server-side LLM calls
+- No database or cache
+- No file writes from the MCP server
+- Local stderr logging only
 
 ## Related
 
-- [`@agent-lint/cli`](https://www.npmjs.com/package/@agent-lint/cli) — CLI interface
-- [Repository](https://gitlab.com/bsamilozturk/agentlint)
+- [Root README](https://github.com/samilytu/agentlint)
+- [CLI package](https://www.npmjs.com/package/@agent-lint/cli)
+- [Authoritative publish source](https://gitlab.com/bsamilozturk/agentlint)
 
 ## License
 

package/dist/bin.d.ts

@@ -0,0 +1,3 @@
+#!/usr/bin/env node
+export {};
+//# sourceMappingURL=bin.d.ts.map

package/dist/bin.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"bin.d.ts","sourceRoot":"","sources":["../src/bin.ts"],"names":[],"mappings":""}

package/dist/bin.js

@@ -4,7 +4,7 @@ import {
   logMcp,
   runHttpServer,
   runStdioServer
-} from "./chunk-MNSQOOMH.js";
+} from "./chunk-DYP46E6V.js";
 
 // src/bin.ts
 var nodeVersion = parseInt(process.versions.node.split(".")[0], 10);

package/dist/catalog.d.ts

@@ -0,0 +1,5 @@
+export declare const CURRENT_TOOL_IDS: readonly ["agentlint_get_guidelines", "agentlint_plan_workspace_autofix", "agentlint_quick_check", "agentlint_emit_maintenance_snippet"];
+export declare const CURRENT_RESOURCE_URIS: readonly ["agentlint://guidelines/{type}", "agentlint://template/{type}", "agentlint://path-hints/{type}"];
+export declare const CURRENT_TOOL_TIMEOUTS: Record<(typeof CURRENT_TOOL_IDS)[number], number>;
+export declare const LEGACY_TOOL_TIMEOUT_ALIASES: Record<string, number>;
+//# sourceMappingURL=catalog.d.ts.map

package/dist/catalog.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"catalog.d.ts","sourceRoot":"","sources":["../src/catalog.ts"],"names":[],"mappings":"AAAA,eAAO,MAAM,gBAAgB,0IAKnB,CAAC;AAEX,eAAO,MAAM,qBAAqB,4GAIxB,CAAC;AAEX,eAAO,MAAM,qBAAqB,EAAE,MAAM,CAAC,CAAC,OAAO,gBAAgB,CAAC,CAAC,MAAM,CAAC,EAAE,MAAM,CAKnF,CAAC;AAEF,eAAO,MAAM,2BAA2B,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAU9D,CAAC"}