Promptless provides visibility into how AI agents interact with websites and APIs, capturing browser actions, network requests, tool calls, and execution traces. Inspect agent workflows, debug failed interactions, analyze behavior, and identify issues in autonomous web automation.
DeepSeek Harness (DSH) is DeepSeek's agent harness. Each MCP server is one plugin entry in your project config.
This guide connects Promptless MCP to DeepSeek Harness through MewCP, so DeepSeek Harness can use Promptless MCP's 20 tools, like submit_documentation_task, request_changes and update_suggestion, without your Promptless MCP credentials ever reaching the app.
Promptless MCP signs in through managed OAuth. MewCP stores the token encrypted, refreshes it before it expires and injects it only at call time. Your agent never holds it, so it can never leak it.
More apps
Manage repositories, pull requests, issues, branches and development workflows on GitHub. Automate software delivery and collaborate across engineering projects.
Connect to external APIs, webhooks and internet services through standard HTTP requests. Build integrations with virtually any platform that exposes a web-based interface.
Give AI coding agents access to a curated library of real-world website designs, screenshots, design systems, and reusable UI components. Search visual references, find similar designs, analyze layouts, extract colors and typography, compare websites, and retrieve copy-ready JSX components for building better interfaces
HotAPI provides a unified API for AI-powered image, video, and audio generation, image editing, and conversational AI. Developers can access multiple models through a single API key, track asynchronous generation tasks, and retrieve results with polling or webhooks.
Other clients
Promptless MCP works the same way in every client below. Pick yours for its own guide.
Free to start.
Scope access to the exact Promptless MCP tools and accounts you allow, work and personal side by side. Anything outside that scope is refused at call time, even when the model asks for it.
Each call is logged: what ran, on which account, and whether it worked. Failures come back structured, so the agent knows whether to retry instead of guessing.
Every app sits behind one secure MCP URL, and there is no server for you to deploy or keep running. Add an integration in the dashboard and it is live on the next call, with no config to change and no new credentials to wire up. Tools load only when a task needs them, so your agent stays fast and focused however many apps you connect.
In your DSH project, or the patch file you mount plugins from.
One plugin instance is one MCP server. Add this to your plugin list.
Tools register as mcp__mewcp__<tool_name>. Sign in when the browser window opens.
Start a new chat and ask: "List the tools you can use from MewCP." You should see your connected apps.
Ask Promptless to update your documentation. Promptless researches the affected docs and proposes changes when needed. Returns a trigger_event_id right away, but the work itself takes several minutes. The response includes the next_call for following the submitted work with wait_for_task_update. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/#submit_documentation_task
Ask Promptless to change a suggestion it already produced, the same as the Request Changes button in the Promptless dashboard. Any changes apply to that suggestion and its existing pull request if one is open. To make the edit yourself instead, edit that pull request directly — its URL is the suggestion's docs_pr_url. Takes a suggestion id from search_suggestions or get_task_status and returns a trigger_event_id for the new task. Use submit_documentation_task when no suggestion exists yet, and when the suggestion is closed or merged and can no longer be changed. The response includes next_call for following the submitted work with wait_for_task_update. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/#request_changes
Change a suggestion's own fields and open its documentation pull request. Set the title, labels, or assignees, and pass open_pull_request to publish the branch as a pull request you can review and edit directly. Every argument is optional except suggestion_id; an omitted field is left alone. Supplying labels or assignees replaces the whole list, so read the current values from search_suggestions or get_task_status first. A new title is also pushed to the suggestion's pull request while that pull request is open, so the two never disagree; if the host rejects the push, the stored title is left unchanged. Opening is idempotent: a suggestion already carrying a pull request returns it rather than opening a second, and draft state follows the organization's source-pull-request policy. Use request_changes to ask Promptless to rewrite the documentation itself. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/#update_suggestion
Close a suggestion and its documentation pull request, ending the review without shipping it. Use it to reject a proposal while triaging search_suggestions results. The pull request is closed on GitHub or GitLab first, so a suggestion whose host refuses the close stays open and the call reports why. Closing cannot be undone from here, and a suggestion that already merged is refused rather than closed. Use request_changes instead when the proposal is worth keeping but needs rework. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/#close_suggestion
List the doc collections Promptless maintains for this organization, with the id to pass as doc_collection_id when submitting a task. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/#list_doc_collections
Follow a Promptless task: wait up to 30 seconds for new messages, changed results, or completion. Omit cursor for an immediate initial snapshot; reuse the returned cursor to receive later messages. A timeout is normal: use next_call to continue waiting without a separate sleep. Drain has_more_messages even when finished is true. Read resolution and suggestions for the outcome. Use get_task_status for a one-time check. Waiting does not cancel cloud work. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/#wait_for_task_update
Check the state of any documentation task in this organization and read the suggestions it produced. It covers tasks you submitted here and tasks that started in Slack, GitHub, or the Promptless dashboard. A finished task can legitimately produce nothing, so read resolution and the suggestions list rather than assuming a pull request exists. This call returns one status snapshot; finished indicates completion, and suggestions can appear while work continues. Each response includes the latest 100 conversation messages, oldest first. Track message IDs to relay new Dispatch messages once, including messages in the final response. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/#get_task_status
Send additional customer instructions to an existing task submitted through the API, MCP, or the dashboard's New Task page, without creating another task. Other sources return unsupported_task_source. Only received or processing tasks accept messages. Finished tasks, including blockers and clarification requests, return task_finished and are not reopened. Reuse an idempotency_key when retrying the same message during active work. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/#send_task_message
List this organization's recent documentation tasks, newest first, each with the trigger_event_id needed to check it. Use it when a task id was not kept, or to see what Promptless has been working on. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/#list_recent_tasks
Search this organization's documentation suggestions by keyword, review status, and label. Use it to check whether a suggestion for a change already exists before submitting a new task, to find a suggestion id to request changes on, to pull a labelled queue such as the P0 suggestions, or to review past suggestions. Results are capped; when truncated is true, narrow the query. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/#search_suggestions
Answer one completed task's marked clarification. Omit answer to ask the user through native input. Supply answer only when the user already provided it; never invent an answer. Acceptance creates a new task. Decline or cancel stops following. If native input is unavailable, ask in chat and call again with the user's answer. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/#answer_task_question
Report this Promptless organization's name, whether a person has claimed it, when an unclaimed agent-created organization expires, which integrations are connected, and the next setup steps. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/
Create a link that lets a person connect one integration to this organization in their browser. Send connect_url to someone who administers that tool. They sign in to Promptless and must be a member of this organization; the first person to open a link for an unclaimed organization claims it and becomes its owner. Call get_account_status after they finish to confirm the connection. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/
Report a Promptless problem or ask the Promptless team for help. Include relevant task or suggestion IDs in concern. Sends a one-way report; no reply comes back through this tool. Do not wait for a reply. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/
List the files in a doc collection's repository, inside the paths Promptless maintains, at the collection's default branch or at a suggestion's branch. Use it to find a page's path before reading or editing it. Results are sorted and capped; when truncated is true, narrow with path_prefix. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/
Read one file's text from a doc collection at its default branch, or at a suggestion's branch to see the proposed version. The path must lie inside the collection's maintained paths. Files over 256 KB and binary files are refused. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/
Find lines matching a substring or an RE2 regular expression (syntax similar to grep -E) across a doc collection's text files (Markdown, MDX, reStructuredText, YAML, JSON, HTML, and similar), at the default branch or at a suggestion's branch. Each match carries the file path, line number, and line text. Use it to find every page that mentions a feature, flag, or endpoint before changing it. The scan reads at most 400 files within a time budget; files_skipped reports how many were left unread, and truncated reports that more lines matched than were returned. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/
Show what a suggestion changes: one unified diff per file against the collection's base branch, plus the head commit to pass as expected_head_sha when editing. Use it to review a suggestion before labelling, editing, or opening its pull request. Diff text is capped in total; when truncated is true, later files carry no diff text. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/
Create or replace one file on an open suggestion's branch with the full content given, as one commit attributed to you. Use it for a new page or a rewrite; use edit_documentation_file for changes to part of a page. The path must lie inside the collection's maintained paths. Open the pull request with update_suggestion afterwards. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/
Replace exact text in one file on an open suggestion's branch, as one commit attributed to you. Each edit's old_text must occur exactly once in the file, so include enough surrounding lines to make it unique; the call is refused, with the match count, when it does not. Edits apply in order. Read the file first so old_text matches it verbatim. Documentation: https://promptless.ai/docs/for-docs/connect/triggers/mcp/
