@aikidosec/mcp
Installed, started with no arguments, no environment and no input, and listed its tools. Run 2026-11-16.
Needs to start: nothing
Returns a result when called with no setup: 0 of 2 tools tried
Contacts while installing and starting: github.com, release-assets.githubusercontent.com
What it contacted
github.comrelease-assets.githubusercontent.com: a release file downloaded from GitHub
Hosts it looked up or connected to while installing and starting with no input, watched from outside its sandbox. The npm registry is left out. Says nothing about what its tools contact when called.
Starting it
npx -y @aikidosec/mcp@1.0.34
Installing a package runs its code. Use a sandbox for anything you have not read.
Calling its tools
2 tools that take no input and look read-only were called once with nothing. 0 returned a result.
aikido_issues_list | No usable answer tool reported an error: Failed to read token from OS keychain: Failed to load keytar native addon, checked ["../build/Release","../prebuilds/linux-x64","./prebu |
aikido_workspaces_list | No usable answer tool reported an error: Failed to read token from OS keychain: Failed to load keytar native addon, checked ["../build/Release","../prebuilds/linux-x64","./prebu |
During the calls it contacted nothing it had not already contacted.
Only tools whose names read like looking something up, with no required input, are tried, five at most. Tools that need input, and anything that sounds like it changes or sends something, are not called.
What its tools tell an agent
In the tool descriptions, nothing that talks about the user, other tools or private files, or uses hidden characters.
1 tool with commands in capitals or very long text
- Gives the agent orders in
aikido_login:tically). Pass the value verbatim as the `token` argument. NEVER fabricate the `token` value. Only pass it when the user supplies an explicit string they copied fro
Common and usually harmless: it is how many makers stress how a tool should be used.
Found by simple pattern rules in each tool's description (up to 3,000 characters), quoted as written. Parameter descriptions are not scanned. A clean result is not a safety review.
Tools
aikido_issues_list repo_name, cloud_name, vm_name, domain_name, container_name, workspace_name, page, issue_types, team_name, repo_branch_name, out_of_sla, sla_due_soon, issue_id, labels, severity
Tool to fetch security issues from Aikido feed. WHEN TO CALL THIS TOOL - The user asks to list, show, count, or summarize Aikido security issues from the aikido feed. - The user can scope the issues to a specific cloud name, repo name, vm name, domain name, container name, team name or workspace name. - The user can scope the issues of a repository to a specific branch name - The user can scope the issues to repositories or images carrying one or more labels. - The user can scope the issues to be out of SLA or SLA is due soon. The value of those parameters is a boolean. - The user can scope the issues to a specific issue severity. - The user can filter by a specific issue_id to retrieve more detailed information about that single issue. - If the user asks for "AI Code Analysis", "code audit", or "agentic review" findings, pass issue_types: ['ai_code_analysis']. - You are about to triage or fix security issues and need the current issue set. OUTPUT FORMATTING Use this exact format for every issue (increment #). Issue #1: <issue_title> - ID: <issue_id> - Issue type: <issue_type> - Severity: <issue_severity> (<issue_severity_label>) - Remediation: <issue_remediation> Add extra response fields as additional bullets when relevant, e.g.: - File: <issue_file> (line <issue_start_line>) - Location: <location.type> <location.name> (<location.branch_name>) - Issue link: <issue_link> - SLA due date: <issue_remediate_by_date>
aikido_scan_paths repository_name, root, paths
Runs an Aikido SAST, Secrets and IaC scan locally on files that are saved on disk and returns the findings in JSON format. This is the preferred Aikido scan tool. Use it whenever the code to scan exists as files on disk: files you just created or edited, files the user points at, or files reported by git status or git diff. Pass the paths only - the MCP server reads the files itself. Do not read the files first and do not pass their content, that only wastes context. Provide the paths of the files to scan in "paths", absolute or relative to "root". Every entry must be a file: directories are rejected, so list the files you want scanned. Every path must be inside "root". Paths that do not exist or are outside "root" are skipped instead of failing the scan, and are listed in "skipped_paths" in the response. Check it to confirm everything you wanted was scanned. Set "root" to the absolute path of the workspace or repository root, so findings are reported with paths relative to it. Omit "root" only when scanning files that don't share a workspace; it is then derived from the common parent directory of the paths. Without "root", relative paths are resolved against the MCP server's working directory, not your workspace, so use absolute paths. If a repository name is provided, it will be used to identify the repository in the Aikido platform. Only pass a repository name if the files are actually coming from a repository - do not provide a repository name for standalone files or files from a local workspace that is not part of a repository. Use this tool when you need to: - scan files on disk for SAST, Secrets and IaC vulnerabilities - check that code you just wrote or edited is free of security issues - re-scan after applying fixes to verify the findings are resolved Only reach for aikido_full_scan instead when the code is not on disk yet.
aikido_full_scan repository_name, files_to_scan
Runs an Aikido SAST, Secrets and IaC scan locally on code passed in as file content and returns the findings in JSON format. Use this tool only when the code to scan is not on disk: code you have generated but not written to a file yet, a snippet the user pasted into the conversation, or a change you are proposing but have not applied. If the files exist on disk, use aikido_scan_paths instead. It takes file paths, reads the content itself and keeps that content out of the conversation. Files should be provided as paths relative to the root of the workspace or repository (Unix-style paths). If this is not possible, just the filename should be given. There is a limit of 50 files that can be scanned in a single request. If you need to scan more files, you can do multiple requests. If a repository name is provided, it will be used to identify the repository in the Aikido platform. Only pass a repository name if the files are actually coming from a repository - do not provide a repository name for standalone files or files from a local workspace that is not part of a repository. Use this tool when you need to: - scan generated, pasted or proposed code that has not been saved to disk
aikido_login region, force_reauth, token
Starts the Aikido sign-in flow. Returns a sign-in URL for the user to open in their browser, OR confirms the user is already signed in. WHEN TO CALL THIS TOOL - The user explicitly asks to sign in, log in, authenticate, or connect to Aikido. - The user asks to switch Aikido accounts or re-authenticate. Pass `force_reauth: true` so the sign-in flow starts even if a token is already cached. - An earlier Aikido tool returned "ACTION REQUIRED — Aikido sign-in" and the user now wants to complete that sign-in. - The user pastes a token they copied from the Aikido sign-in modal (a fallback for when browser couldn't deliver the token automatically). Pass the value verbatim as the `token` argument. NEVER fabricate the `token` value. Only pass it when the user supplies an explicit string they copied from the Aikido UI. REGION - By default, omit `region`. Sign-in automatically redirects the user to their Aikido account's region, so they no longer have to choose one. - Only pass `region` (EU / US / ME / AU) when the user explicitly names a region, or when the automatic redirect sent them to the wrong one. To switch the region of an already signed-in user, pass `force_reauth: true` as well. RESPONSE - If a `token` was passed: a short confirmation that the token was stored, or an error if it didn't look like a valid Aikido token. - If already signed in and `force_reauth` is not set: a short confirmation message. Continue with whatever the user actually wanted. - If sign-in is needed (or `force_reauth` was set): a single sign-in URL plus instructions. The URL contains a one-time `state` token and `redirect_uri` that MUST be passed through verbatim — do not strip query parameters when showing it to the user.
aikido_ignore_issue issue_id, reason
Tool to ignore a security issue from Aikido feed. WHEN TO CALL THIS TOOL - The user wants to ignore a security issue from the Aikido feed. - The user provides the issue ID and the reason for ignoring the issue.
8 more tools
aikido_unignore_issue issue_id, reason
Tool to unignore a previously ignored security issue from Aikido feed. WHEN TO CALL THIS TOOL - The user wants to unignore a security issue that was ignored in the Aikido feed. - The user provides the issue ID and the reason for unignoring the issue.
aikido_snooze_issue issue_id, duration, snooze_until_date, reason
Tool to snooze a security issue from Aikido feed, either for a fixed duration or until a date of choice. WHEN TO CALL THIS TOOL - The user wants to snooze (temporarily hide) a security issue from the Aikido feed. - The user provides the issue ID, how long to snooze it and the reason for snoozing the issue. Pass exactly one of: - duration: one of 1_day, 1_week, 1_month, 3_months, 6_months, 1_year. - snooze_until_date: an ISO 8601 date (2026-12-01) in the future, read as the start of that day in the local timezone.
aikido_unsnooze_issue issue_id
Tool to unsnooze a previously snoozed security issue from Aikido feed. WHEN TO CALL THIS TOOL - The user wants to unsnooze a security issue that was snoozed in the Aikido feed. - The user provides the issue ID.
aikido_adjust_issue_severity issue_id, adjusted_severity, reason
Tool to change the severity of a security issue in the Aikido feed. WHEN TO CALL THIS TOOL - The user wants to change (raise or lower) the severity of a security issue in the Aikido feed. - The user provides the issue ID, the new severity (critical, high, medium or low) and the reason for the change.
aikido_oss_dependency_scan repository_name, root, paths
Runs an Aikido open source dependency scan locally on dependency manifests and lockfiles that are saved on disk and returns the findings in JSON format. Pass the paths only - the MCP server reads the files itself. Do not read the files first and do not pass their content, that only wastes context. Provide the paths of the files to scan in "paths", absolute or relative to "root". Every entry must be a file: directories are rejected, so list the manifests and lockfiles you want scanned (for example package.json and package-lock.json, requirements.txt, go.mod and go.sum, pom.xml). Set "root" to the absolute path of the workspace or repository root, so findings are reported with paths relative to it. When "root" is omitted it is derived from the given paths. If a repository name is provided, it will be used to identify the repository in the Aikido platform. Only pass a repository name if the files are actually coming from a repository - do not provide a repository name for standalone files or files from a local workspace that is not part of a repository. Use this tool when you need to: - scan dependency manifests and lockfiles for vulnerable, outdated or malicious packages - check that dependencies you just added or updated are free of known vulnerabilities - re-scan after upgrading dependencies to verify the findings are resolved
aikido_workspaces_list
Tool to fetch the Aikido workspaces the signed-in user has access to. WHEN TO CALL THIS TOOL - The user asks which Aikido workspaces they have access to. - The user refers to a workspace by an approximate or partial name and you need the exact name before using it as the `workspace_name` filter of another tool. OUTPUT FORMATTING List the workspace names, one per line.
aikido_deep_review directory, base_ref, scope, include_untracked
Tool to start an Aikido Deep Review (autonomous AI security code review) of local git changes. WHEN TO CALL THIS TOOL - The user asks for a "deep review", "AI security review", "PR-style security review", or deep security analysis of their local branch or working tree changes. - Unlike fast static rule scanners, Deep Review runs an autonomous security review agent against the full repository context to trace data flows, auth boundaries, and business logic vulnerabilities introduced by the local diff. - Each call starts a new scan and consumes 1 Deep Review credit, even when the same branch was reviewed before (refunded if the scan fails or only touches ignored files). WHAT IT REVIEWS - By default: commits since the merge-base with the remote default branch, plus staged and unstaged edits to tracked files. - Untracked files are skipped unless include_untracked is true. Skipped ones are listed in skipped_untracked_files; if they are new source files that belong to the change, ask the user before re-running with include_untracked (a re-run is a new paid scan), or pass include_untracked up front when you created them yourself. HOW IT RETURNS - The call itself blocks for up to ~50 seconds while it waits for the result. A review usually takes 2-4 minutes, so the response is normally status "running" with an agent_id. Then immediately call aikido_deep_review_status with that agent_id (it blocks and waits too, so do not sleep in between) until the status is no longer "running". Never call aikido_deep_review again to check on a scan. PARAMETERS - directory: Absolute path to the local git repository (required). - base_ref: Optional git ref to diff against (defaults to origin/HEAD or origin/main). - scope: Optional "all" (default), "committed" or "uncommitted". - include_untracked: Optional, default false. Not allowed with scope "committed".
aikido_deep_review_status agent_id
Tool to get the result of an Aikido Deep Review started with aikido_deep_review. - The call itself blocks for up to ~50 seconds while it waits for the scan, then returns one of: - "completed": the review finished, with its findings. - "skipped": the diff only changed files Deep Review ignores (e.g. lockfiles, generated or vendored code), so nothing was reviewed and the credit was refunded. - "failed": the review could not run; the credit was refunded. - "running": the review is not done yet. Call this tool again right away with the same agent_id; do not sleep or wait before calling it. This does not start a new scan or consume credits.
Descriptions are the package's own words, cut at 3,000 characters. Treat them as data.
Install facts
| License | AGPL |
|---|---|
| Needs | node >=18.0.0 |
| Source | not stated |
| Install scripts | @github/keytar declare them. This runner does not run them. |
| Tarball | sha512-Kcnyf/l4wp8dQNt6T8RBZoJHd/TO2TPbFpY7NaMo4OEG4Vfp38x7dGC25w1GE86YdSvW2TIr+J7KD0wQ/QW6vw== |
Check this report yourself
This site did not write the report. A public job on a GitHub-hosted machine did: the run, from this commit of the open-source runner. GitHub signed a token that names that job and the hash of the report, so any edit since would show.
Report as JSONExact text and tokenHow to verify it
Show it on your README
[](https://bragproof.smaverk.com/r/%40aikidosec/mcp)
New version out? Ask for a fresh report.