// SCAN REPORT

Promethean OS MCP

69 tools. 50 can modify or destroy data without limits.

4 destructive tools with no built-in limits. Policy required.

Last updated: 4 Jul 2026

50 can modify or destroy data

19 read-only

69 tools total

Community server · catalogue entry verified 04/07/2026

How to control Promethean OS MCP ↓

TOOL MAP

What Promethean OS MCP exposes to your agents

Read (19) Write / Execute (45) Destructive / Financial (4)

Running more than one server? Check your whole stack →

MOST DANGEROUS TOOLS

Critical Risk

The most dangerous Promethean OS MCP tools

Destructive · High ollama_remove_job The verb 'remove' combined with the conditional 'if not running' indicates irreversible deletion of a job. Destructive · High sandbox_delete This tool permanently deletes a git worktree sandbox, which is an irreversible operation. Destructive · High github_review_revert_commits Git revert creates new commits that undo previous commits, permanently altering the repository's commit history in a way that is difficult to recover from without additional intervention. Destructive · High pnpm_remove Remove dependencies via pnpm, optionally scoped to workspace filters.

50 of Promethean OS MCP's 69 tools can modify, destroy, or commit something on every call — and an agent calls them with no built-in limits.

POLICY

How to control Promethean OS MCP

PolicyLayer is an MCP gateway — it sits between your AI agents and Promethean OS MCP, and nothing reaches the server without passing your rules. These are the rules we recommend:

Deny destructive operations

{
  "github_review_revert_commits": {
    "deny_if": [
      {
        "conditions": [],
        "on_deny": "Blocked by default. Requires approval."
      }
    ]
  }
}

Destructive tools should never be available to autonomous agents without human approval.

Rate limit write operations

{
  "apply_patch": {
    "limits": [
      {
        "counter": "apply_patch_per_hour",
        "window": "hour",
        "max": 30,
        "scope": "grant"
      }
    ]
  }
}

Prevents bulk unintended modifications from agents caught in loops.

Cap read operations

{
  "discord_list_messages": {
    "limits": [
      {
        "counter": "discord_list_messages_per_minute",
        "window": "minute",
        "max": 60,
        "scope": "grant"
      }
    ]
  }
}

Controls API costs and prevents retry loops from exhausting upstream rate limits.

Create a free account and register Promethean OS MCP — nothing to install.
Add these rules — paste them, or build them visually. Tune the limits to your setup.
Point your MCP client (Claude, Cursor, anything) at your gateway URL.

ENFORCE POLICY ON PROMETHEAN OS →

Instant setup, no code required.

FULL CATALOGUE

All 69 Promethean OS MCP tools

DESTRUCTIVE 4 tools

Destructive github_review_revert_commits Revert one or more commits using git revert. Destructive ollama_remove_job Remove a job by id or name if not running Destructive pnpm_remove Remove dependencies via pnpm, optionally scoped to workspace filters. Destructive sandbox_delete Remove a git worktree sandbox by sandboxId.

EXECUTE 20 tools

Execute exec_run Execute a vetted shell command from the allowlist. Supports optional args when explicitly enabled. Execute github_graphql Post a GraphQL query to GitHub. Execute github_request Call GitHub REST API with optional ETag cache & pagination. Execute github_review_checkout_branch Check out an existing git branch. Execute github_review_push Push the current branch to the specified remote with optional upstream/force settings. Execute nx_generate_package Generate a new workspace package using the Nx tools:package generator. Execute ollama_enqueue_chat_completion Queue a chat completion against a conversation or raw messages Execute ollama_enqueue_job_from_template Queue a job that will execute a named template Execute ollama_pull Queue a model pull (no-op executor in MVP) Execute ollama_start_conversation Create a conversation; optionally seeds initial user message Execute pnpm_install Run pnpm install at the workspace root or limited to filtered packages. Execute pnpm_run_script Execute a pnpm script, optionally filtered to specific workspace packages. Execute process_enqueue_task Enqueue a command for execution respecting concurrency limits. Execute process_stop Stop a running task via id, pid, or name and return trailing output. Execute tdd_coverage Run c8+AVA to produce coverage summary; enforce optional thresholds. Execute tdd_mutation_score Run Stryker mutation testing and return the score (fail if below minScore). Execute tdd_property_check Dynamically import a module export that builds a fast-check property and assert it. Execute tdd_run_tests Run AVA via npx with JSON (or TAP) output and return aggregated results. For long-running watchers, use tdd_st Execute tdd_start_watch Start an AVA --watch process and stream output via getWatchChanges. Execute tdd_stop_watch Stop the running watch process.

WRITE 25 tools

Write apply_patch Apply a universal diff patch to the MCP sandbox using git apply. Write discord_send_message Send a message to a Discord channel. Provide content or embeds and the Discord space URN. Write files_write_content Write UTF-8 content to a file (creates if not exists). Write files_write_lines Append or insert lines into a file at startLine (1-based). Write github_apply_patch Apply a unified diff to a GitHub branch by committing the changes via createCommitOnBranch. Write github_contents_write Create or update a file via the GitHub contents API with automatic base64 encoding. Write github_pr_review_comment_inline Create an inline review thread on a pull request with automatic diff resolution. Write github_pr_review_start Create a pending review on a pull request via GraphQL. Write github_pr_review_submit Submit a pending review with APPROVE, REQUEST_CHANGES, or COMMENT. Write github_review_commit Create a git commit. Optionally stage specific paths or use --all/--allow-empty. Write github_review_create_branch Create and check out a new git branch optionally from a specific start point. Write github_review_open_pull_request Open a new pull request targeting a base branch with the provided title and body. Write github_review_request_changes_from_codex Request changes from Codex by posting an issue-level pull request comment tagging @codex. Write github_review_submit_comment Submit an issue-level comment on a pull request. Write github_review_submit_review Create a pull request review with optional summary body and inline comments. Write mcp_toolset Describe this endpoint Write ollama_create_template Create or update a template (s-expr text stored verbatim) Write ollama_enqueue_generate_job Queue a text generation job (no execution in MVP) Write pnpm_add Add dependencies via pnpm, optionally scoped to specific workspace packages. Write pr.enableAutoMerge Enable auto-merge for a clean PR. Write pr.enqueue Add PR to merge queue. Write pr.updateBranch Server-side merge base→head (like the Update branch button). Write process_update_task_runner_config Update a single key in the task runner configuration. Write sandbox_create Create a git worktree-based sandbox rooted under .sandboxes/<id>. Write tdd_scaffold_test Create or append a test file next to a module (unit or property-test template).

READ 19 tools

Read discord_list_messages List recent messages from a Discord channel with optional pagination parameters. Read files_list_directory List files and directories within the sandbox root. Read files_search Search file contents under a directory and return matching line snippets. Read files_tree_directory Build a tree-like view of a directory, with optional hidden files and max depth. Read files_view_file View a file by path, with line-context selection. Read github_pr_files List files in a pull request with parsed diff hunks for line resolution. Read github_pr_get Fetch metadata for a pull request (ids, SHAs, author). Read github_review_get_action_status Fetch the latest workflow and check status for the most recent pull request commit. Read github_review_get_comments List issue comments on a pull request with pagination support. Read github_review_get_review_comments Fetch review thread comments (diff comments) for a pull request. Read github_workflow_get_job_logs Download and extract the log files for a GitHub Actions workflow job. Read github_workflow_get_run_logs Download and extract the log files for a GitHub Actions workflow run. Read mcp_endpoints List all configured endpoints with metadata and includeHelp flag. Read mcp_help List available tools with args, defaults, outputs, and examples. Read mcp_validate_config Validate endpoint/tool configuration and narrative metadata. Returns errors and warnings. Read pr.status Get PR mergeability and state; polls until not UNKNOWN if requested. Read sandbox_list List sandboxes created as git worktrees under .sandboxes. Read tdd_changed_files List files changed vs a git base ref, filtered by glob patterns. Read tdd_get_watch_changes Get incremental stdout/stderr from the running watch.

OTHER 1 tools

Other test_tool A simple test tool

EXPLORE

Related servers

Other MCP servers with similar tools — same risk classification, starter policies for each.

FAQ

Questions about Promethean OS MCP

Can an AI agent delete data through the Promethean OS MCP server? +

Yes. The Promethean OS MCP server exposes 4 destructive tools including github_review_revert_commits, ollama_remove_job, pnpm_remove. These permanently remove resources with no undo. PolicyLayer blocks destructive tools by default so they never reach the upstream server.

How do I prevent bulk modifications through Promethean OS MCP? +

The Promethean OS MCP server has 25 write tools including apply_patch, discord_send_message, files_write_content. Set a rate limit in your policy -- for example, 10 calls per hour prevents an agent from making more than 10 modifications per hour. PolicyLayer enforces this at the gateway, before calls reach Promethean OS MCP.

How many tools does the Promethean OS MCP server expose? +

69 tools across 4 categories: Destructive, Execute, Read, Write. 19 are read-only. 50 can modify, create, or delete data.

How do I enforce a policy on Promethean OS MCP? +

Register the Promethean OS MCP server in PolicyLayer, apply the suggested rules above (adjust the limits to your use case), and point your AI client at the PolicyLayer proxy URL instead of the server directly. Your agents keep the same tools; PolicyLayer evaluates every call against policy before it executes. Nothing to install, live in minutes.

Enforce policy on every Promethean OS MCP tool call.

Deterministic rules across all 69 Promethean OS MCP tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.

ENFORCE POLICY ON PROMETHEAN OS →

Instant setup, no code required.

69 Promethean OS MCP tools catalogued and risk-classified — across an index of 43,000+ MCP servers.