chore(github-tag): pin tag devsecninja/ai-toolkit to 2491ced [automerge] - #339

🏁 Script executed:

#!/bin/bash
set -eu
printf '%s\n' '--- license file ---'
cat -n .agents/skills/tripit-exporter/LICENSE
printf '%s\n' '--- nearby files ---'
find .agents/skills/tripit-exporter -maxdepth 2 -type f -print | sort
printf '%s\n' '--- relevant references ---'
rg -n -i --hidden --glob '!node_modules' --glob '!dist' 'Casey Grippi|tripit-exporter|TripIt exporter|Personal use|redistribut' .agents README.md .github 2>/dev/null || true

Repository: DevSecNinja/.github

Length of output: 1612

🌐 Web query:

https://lobehub.com/skills/caseyg-caseys-claude-tripit-export license redistribution permission

💡 Result:

<search_synthesis>
The plugin identifier caseyg-caseys-claude-tripit-export refers to a specific skill hosted in the LobeHub Skills Marketplace [1]. Because this is a user-contributed plugin hosted on a third-party repository (caseyg/caseys-claude) [2][1], its license is determined by the author of that repository, not by the LobeHub platform&#39;s license itself [3][4]. To determine the redistribution and usage permissions for this specific plugin, you must consult the repository where the source code is hosted: 1. Visit the repository: https://github.com/caseyg/caseys-claude [1] 2. Look for a LICENSE file in the root of the repository or the plugin&#39;s directory. 3. If no explicit license is provided, copyright law defaults to "all rights reserved," meaning you do not have permission to redistribute or modify the code without express authorization from the author, regardless of its availability on the LobeHub marketplace. Note that while LobeHub itself uses a custom license (the LobeHub Community License) for its core software [3][5], this does not automatically apply to third-party plugins submitted to their marketplace. The LobeHub platform acts as a directory; it does not grant you rights to the intellectual property of individual plugin developers [1]. Always check the repository&#39;s license file before redistributing or using the code commercially.
</search_synthesis>

<source_evidence>

Citations:

• 1: https://lobehub.com/skills/caseyg-caseys-claude-tripit-export
• 2: https://claudemarketplaces.com/plugins/caseyg-caseys-claude/tripit-export
• 3: https://github.com/lobehub/lobe-chat/blob/next/LICENSE
• 4: https://lobehub.com/blog/towards-lobe-chat-v1
• 5: GitHub issue 9325 in lobehub/lobehub (link omitted to avoid creating a cross-reference)

🌐 Web query:

site:github.com/caseyg/caseys-claude LICENSE tripit-export redistribution permission Casey Grippi

💡 Result:

<search_synthesis>
There is no publicly available information regarding specific redistribution permissions or an explicit open-source license for the tripit-export tool developed by Casey Gollan (identified in search results as caseyg) [1][2]. The tool is hosted within a repository titled caseys-claude, which is described as containing personal skills and configuration files intended for personal use, with the author noting that these skills often include hardcoded assumptions that require adaptation for other users [1]. Because no license is provided in the repository, you should assume the code is under standard copyright, meaning it cannot be redistributed or modified without explicit permission from the author [1][2].
</search_synthesis>

<source_evidence>

Citations:

• 1: https://github.com/caseyg/caseys-claude
• 2: https://github.com/caseyg

🏁 Script executed:

#!/bin/bash
set -eu
printf '%s\n' '--- skill outline/context ---'
wc -l .agents/skills/tripit-exporter/SKILL.md
sed -n '1,230p' .agents/skills/tripit-exporter/SKILL.md
printf '%s\n' '--- repository language/config references ---'
rg -n --hidden --glob '!node_modules' --glob '!dist' --glob '!build' \
  'tripit-exporter|SKILL\.md|noImplicitAny|strict|tsconfig|typescript|\.ts\b|\.js\b' \
  .agents package.json package-lock.json pnpm-lock.yaml yarn.lock tsconfig.json jsconfig.json 2>/dev/null || true

Repository: DevSecNinja/.github

Length of output: 6697

🏁 Script executed:

#!/bin/bash
set -e
printf '%s\n' '--- script outline ---'
ast-grep outline .github/hooks/scripts/ai-toolkit/scripts/tool-guardian/guard-tool.sh 2>/dev/null || true
printf '%s\n' '--- script relevant sections ---'
cat -n .github/hooks/scripts/ai-toolkit/scripts/tool-guardian/guard-tool.sh | sed -n '1,180p'
printf '%s\n' '--- related hook/config references ---'
rg -n -i --hidden --glob '!node_modules' --glob '!dist' --glob '!build' 'toolName|disableAllHooks|preToolUse|guard-tool|json_escape' .github README.md 2>/dev/null | head -240

Repository: DevSecNinja/.github

Length of output: 10772

🏁 Script executed:

#!/bin/bash
set -e
printf '%s\n' '--- remaining script ---'
cat -n .github/hooks/scripts/ai-toolkit/scripts/tool-guardian/guard-tool.sh | sed -n '179,260p'
printf '%s\n' '--- hook configuration ---'
cat -n .github/hooks/ai-toolkit-tool-guardian.json
printf '%s\n' '--- repository guidance and schemas mentioning hook input ---'
rg -n -i --hidden --glob '!node_modules' --glob '!dist' --glob '!build' 'toolInput|toolName|preToolUse|hook.*schema|schema.*hook|hook.*input|input.*hook' . 2>/dev/null | head -300

Repository: DevSecNinja/.github

Length of output: 2666

🌐 Web query:

official GitHub Copilot hooks input schema toolName control characters preToolUse

💡 Result:

<search_synthesis>
The preToolUse hook in GitHub Copilot is a powerful mechanism that allows developers to intercept, inspect, and modify tool calls before they are executed by the agent [1]. Input Schema The input received by the preToolUse hook depends on how it is configured: 1. Standard Configuration: When using the standard hook format, the input provides context about the tool invocation [2][3][4]. Key fields include: - toolName (string): The identifier of the tool being called (e.g., bash, read_file) [2][4]. - toolArgs (object or JSON string): The arguments provided to the tool [2][4]. Note that in some environments (like the Copilot CLI), toolArgs is passed as a JSON string and must be explicitly parsed [4]. - timestamp (number): A Unix timestamp of the hook trigger [2]. - cwd (string): The current working directory [2]. 2. Claude-format / PascalCase PreToolUse: When configured with the PreToolUse event name (used for Claude Code plugins or Open Plugins compatibility), the schema uses snake_case fields and reports tool_name using Claude-specific mapping (e.g., Bash instead of bash) [5][6]. The payload includes hook_event_name, session_id, timestamp (ISO 8601), cwd, tool_name, and tool_input [5][6]. Decision Control The preToolUse hook provides explicit control over tool execution. By returning a structured response, you can: - Allow execution: Return { "permissionDecision": "allow" } [2][3]. - Deny execution: Return { "permissionDecision": "deny", "permissionDecisionReason": "..." } [2][3]. - Modify arguments: Include a modifiedArgs field in your response to update the arguments before the tool runs [2][3]. Regarding control characters or specific constraints, the documentation highlights that toolArgs are intended for structured data, and improper JSON formatting in the arguments can lead to parsing errors [4]. When building tools, developers define their parameters using JSON Schema (or Zod schemas in the SDK) to ensure inputs are validated against the expected schema before the hook even processes the call [7][8][9]. Security best practices involve using preToolUse to implement allow-lists for directories or to redact sensitive information from arguments before logging [2][3][4].
</search_synthesis>

<source_evidence>

Citations:

• 1: https://docs.github.com/en/copilot/concepts/agents/hooks
• 2: https://docs.github.com/en/copilot/how-tos/copilot-sdk/hooks/pre-tool-use
• 3: https://github.com/github/copilot-sdk/blob/main/docs/hooks/pre-tool-use.md
• 4: https://docs.github.com/en/copilot/tutorials/copilot-cli-hooks
• 5: https://docs.github.com/en/copilot/reference/hooks-reference
• 6: https://github.com/github/docs/blob/main/content/copilot/reference/hooks-reference.md
• 7: https://github.com/microsoft/vscode/blob/main/extensions/copilot/docs/tools.md
• 8: https://github.com/github/copilot-sdk/blob/584a239e/nodejs/src/types.ts
• 9: https://github.com/github/copilot-sdk/blob/584a239e/java/src/main/java/com/github/copilot/rpc/ToolDefinition.java

🏁 Script executed:

set -eu
printf '%s\n' '--- target file ---'
if [ -f .github/mcp.json ]; then
  cat -n .github/mcp.json
else
  printf '%s\n' '.github/mcp.json not found'
fi
printf '%s\n' '--- nearby MCP references ---'
rg -n --glob '!node_modules' --glob '!dist' --glob '!build' 'mcp\.json|api\.githubcopilot\.com/mcp|X-MCP-Readonly|github-mcp-server' .github .vscode README.md 2>/dev/null || true

Repository: DevSecNinja/.github

Length of output: 967

🌐 Web query:

GitHub Copilot custom MCP servers api.githubcopilot.com/mcp read-only URL X-MCP-Readonly tools wildcard official documentation

💡 Result:

<search_synthesis>
The GitHub MCP Server, maintained by GitHub, allows you to configure remote connections through api.githubcopilot.com/mcp [1][2]. You can customize tool access, enable read-only modes, and filter specific tools using headers, URL paths, or environment variables [1][3]. Read-Only Configuration To restrict the GitHub MCP server to read-only operations, you can use the X-MCP-Readonly header or a specialized URL path: - Header: Set the X-MCP-Readonly header to "true" [3][4]. - URL: Append /readonly to the base URL (e.g., https://api.githubcopilot.com/mcp/readonly) [3][5]. Tool Selection and Wildcards When configuring your MCP server, you can manage which tools are exposed [1][3]: - Wildcards: You can use "" in the tools list to enable all available tools for a given configuration [6][7]. - Individual Tools: You can explicitly list tools (e.g., using the X-MCP-Tools header) to include only specific functionality, such as get_file_contents or issue_read [1][3]. - Toolsets: You can use the X-MCP-Toolsets header to enable collections of tools, such as repos, issues, or pull_requests [1][3]. Configuration Examples For HTTP-based remote configurations in your mcp.json file, you can combine these settings [6][3]: { "mcpServers": { "github-mcp-server": { "type": "http", "url": "https://api.githubcopilot.com/mcp/readonly", "tools": [""], "headers": { "X-MCP-Toolsets": "repos,issues,pull_requests" } } } } In this configuration, read-only mode is enforced via the URL, and the asterisk (*) ensures all tools within the specified toolsets are made available [6][3]. The server treats these settings as strict filters, removing write-capable tools from the list entirely when read-only mode is active [5].
</search_synthesis>

<source_evidence>

Citations:

• 1: https://github.com/github/github-mcp-server/blob/HEAD/docs/server-configuration.md
• 2: https://docs.github.com/en/copilot/how-tos/provide-context/use-mcp-in-your-ide/set-up-the-github-mcp-server
• 3: https://github.com/github/github-mcp-server/blob/main/docs/server-configuration.md
• 4: https://github.blog/ai-and-ml/generative-ai/a-practical-guide-on-how-to-use-the-github-mcp-server/
• 5: https://agenticcontrolplane.com/mcp-controls/github
• 6: https://docs.github.com/en/copilot/how-tos/copilot-on-github/customize-copilot/configure-mcp-servers
• 7: https://docs.github.com/en/copilot/how-tos/copilot-sdk/features/mcp

‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.