{"id":"chu2bard-termlink","name":"termlink","homepage":null,"repo_url":"https://github.com/chu2bard/termlink","category":"devtools","subcategories":[],"tags":["mcp","model-context-protocol","terminal","shell","filesystem","process-info","typescript","developer-tools","automation"],"what_it_does":"termLink is an MCP server that exposes local terminal/shell execution, filesystem operations (read/write/list), directory existence checks, and process/environment introspection (e.g., process info, list processes, get_env) to an MCP client such as Claude Desktop.","use_cases":["Building an agent workflow that can run shell commands","Letting an agent inspect and edit files in a workspace","Debugging/helpdesk automation that needs process/env visibility","Local automation where an operator manually controls what commands the agent can run"],"not_for":["Untrusted or internet-facing deployments without strong sandboxing","Use cases requiring strict least-privilege or strong auditability","Running as-is in multi-tenant environments","Any scenario where exposing filesystem + command execution to an agent is unacceptable"],"best_when":"You control the MCP client, run locally (or in a tightly sandboxed environment), and use configuration to restrict destructive commands.","avoid_when":"You cannot guarantee agent/client trust, or you cannot enforce command/file access constraints beyond the provided config.","alternatives":["Other MCP/local tool bridges with stronger sandboxing policies","Agent frameworks that separate execution from tool exposure (e.g., a job runner/queue)","Containerized execution (e.g., running commands inside a dedicated container with capped permissions)"],"af_score":47.5,"security_score":22.0,"reliability_score":20.0,"package_type":"mcp_server","discovery_source":["github"],"priority":"high","status":"evaluated","version_evaluated":null,"last_evaluated":"2026-03-30T15:34:42.923540+00:00","interface":{"has_rest_api":false,"has_graphql":false,"has_grpc":false,"has_mcp_server":true,"mcp_server_url":null,"has_sdk":false,"sdk_languages":[],"openapi_spec_url":null,"webhooks":false},"auth":{"methods":["None specified (local MCP server started via command)"],"oauth":false,"scopes":false,"notes":"No authentication mechanism is described in the provided README. Access appears to be controlled primarily by local network/process access and whatever constraints are enforced by the MCP server itself (e.g., blockedCommands/allowedCommands)."},"pricing":{"model":null,"free_tier_exists":false,"free_tier_limits":null,"paid_tiers":[],"requires_credit_card":false,"estimated_workload_costs":null,"notes":"No pricing information provided (MIT repo; appears to be self-hosted software)."},"requirements":{"requires_signup":false,"requires_credit_card":false,"domain_verification":false,"data_residency":[],"compliance":[],"min_contract":null},"agent_readiness":{"af_score":47.5,"security_score":22.0,"reliability_score":20.0,"mcp_server_quality":55.0,"documentation_accuracy":55.0,"error_message_quality":null,"error_message_notes":"README contains a note to 'fixme: handle errors' and no documented error formats/codes. Likely uneven/insufficient error reporting for agent reliability.","auth_complexity":100.0,"rate_limit_clarity":0.0,"tls_enforcement":0.0,"auth_strength":10.0,"scope_granularity":40.0,"dependency_hygiene":50.0,"secret_handling":20.0,"security_notes":"Major risk: exposing shell command execution and filesystem/process info to an MCP client. README shows a config with blockedCommands (e.g., rm -rf /) and allowedCommands, but no evidence of sandboxing, privilege dropping, or robust auditing. No auth mechanism is described, so access control likely depends on local trust/network isolation. Secret handling and structured redaction are not documented; agents could potentially read env/process details containing secrets.","uptime_documented":0.0,"version_stability":30.0,"breaking_changes_history":30.0,"error_recovery":20.0,"idempotency_support":"false","idempotency_notes":"No idempotency guidance is provided. Tools include write_file and run_command; repeated calls may have side effects.","pagination_style":"none","retry_guidance_documented":false,"known_agent_gotchas":["Tooling includes direct command execution and file writes; agents may inadvertently run destructive commands if allowed/blocked controls are misconfigured.","Configuration example suggests blockedCommands and allowedCommands, but 'allowedCommands' is shown empty; misconfiguration could result in either overly permissive or overly restrictive behavior depending on implementation.","Potential output truncation/limits via maxOutputSize (configured), which may cause agents to miss context when commands generate more output than allowed."]}}