Skill
tendem-result
Explicitly fetch and present a completed Tendem task's result (markdown + files). Invoke as $tendem-result [task_id] — omit the…
@tolokaInstall SkillSales
Skill
/tendem-task
Explicitly submit a new task to Tendem (hybrid AI + human experts) and drive it through scoping. Invoke as $tendem-task <what you want a human expert to do>.
Install
$ npx -y skills add toloka/tendem-mcp --skill tendem-task --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
- Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
- You can call itInvoke it directly when you want it.
- Slash command
/tendem-task
Context preview
The summary Claude sees to decide when to auto-load this skill.
Explicitly submit a new task to Tendem (hybrid AI + human experts) and drive it through scoping. Invoke as $tendem-task <what you want a human expert to do>.
SKILL.md
tendem-task.SKILL.mdname: tendem-task
description: Explicitly submit a new task to Tendem (hybrid AI + human experts) and drive it through scoping. Invoke as $tendem-task <what you want a human expert to do>.
Submit a new Tendem task from the user's instruction — the text accompanying this invocation.
Follow the `tendem-tasks` skill. Fidelity first:
1. Call `create_task` **now** — `description` is the user's instruction, passed as faithfully as possible. Add only context the user actually stated (input files by name, a deadline if one was given). Do NOT expand it into a synthesized brief, and do NOT pre-ask the user clarifying questions — scoping is Tendem's job. Derive a short `name` (≤120 chars) and reuse a stable `conversation_id` for this conversation. 2. If the instruction references local files: **⚠️ do not upload them yourself** — OpenAI's security policies block the sandbox from PUTting files to external storage (an OpenAI platform restriction, not a Tendem one). Immediately after creating the task, call `get_file_upload_url`, build fully-substituted `curl` upload commands per the `tendem-tasks` skill's `references/files.md`, and hand them to the user to run on their host machine — telling them the manual step is due to OpenAI's sandbox policy. After the user confirms, send the naming `send_message`. 3. Scoping loop: poll with `get_task(task_id, wait_for_change_seconds=30)`, answer Tendem's questions from conversation context via `send_message`, escalate to the user only when the answer isn't in context or scope/deliverables/deadline are at stake. 4. Stop when the quote is ready (`next_action=await_user_approval`): call `get_contract(task_id)` for the full scope + price, show the user both, and wait for their explicit go-ahead before `approve_task`.
Report the `task_id` in your reply so it can be referenced later.
Read more
name: tendem-task description: Explicitly submit a new task to Tendem (hybrid AI + human experts) and drive it through scoping. Invoke as $tendem-task <what you want a human expert to do>.
Submit a new Tendem task from the user's instruction — the text accompanying this invocation.
Follow the `tendem-tasks` skill. Fidelity first:
1. Call `create_task` **now** — `description` is the user's instruction, passed as faithfully as possible. Add only context the user actually stated (input files by name, a deadline if one was given). Do NOT expand it into a synthesized brief, and do NOT pre-ask the user clarifying questions — scoping is Tendem's job. Derive a short `name` (≤120 chars) and reuse a stable `conversation_id` for this conversation. 2. If the instruction references local files: **⚠️ do not upload them yourself** — OpenAI's security policies block the sandbox from PUTting files to external storage (an OpenAI platform restriction, not a Tendem one). Immediately after creating the task, call `get_file_upload_url`, build fully-substituted `curl` upload commands per the `tendem-tasks` skill's `references/files.md`, and hand them to the user to run on their host machine — telling them the manual step is due to OpenAI's sandbox policy. After the user confirms, send the naming `send_message`. 3. Scoping loop: poll with `get_task(task_id, wait_for_change_seconds=30)`, answer Tendem's questions from conversation context via `send_message`, escalate to the user only when the answer isn't in context or scope/deliverables/deadline are at stake. 4. Stop when the quote is ready (`next_action=await_user_approval`): call `get_contract(task_id)` for the full scope + price, show the user both, and wait for their explicit go-ahead before `approve_task`.
Report the `task_id` in your reply so it can be referenced later.
Other skills on tendem-mcp.
@tolokaInstall SkillSkill
tendem-status
Explicitly check on a Tendem task and advance it (poll cleanly, don't busy-loop). Invoke as $tendem-status [task_id] — omit the…
@tolokaInstall SkillSkill
tendem-tasks
Use when a task is non-trivial and could benefit from human involvement — delegating work to a human expert (research, review,…
@tolokaInstall Skill