Skip to main content
Glama

feedback_draft

Draft user-approved feedback about AI Team OS as a GitHub Issue for bugs or a Discussion for ideas, then return the title and body for review.

Instructions

Draft feedback to the AI Team OS maintainers: a bug as a GitHub Issue, an idea in Discussions (Ideas).

Only for AI Team OS itself (its hooks, MCP tools, API, Dashboard, installer), and only after the user said they want to send feedback; feedback about Claude Code, Codex or a model goes to their own channels. You may offer it when you hit an OS error. Decide the kind with the user: "broken" (expected one thing, got another) is bug; "I want" is idea; if a remark could be either, ask. A remark with both: two drafts. Security problems: kind="security" (returns the private reporting link, files nothing). Usage questions: kind="question". Before drafting, search the direction memories for private-term rules and keep such terms out. Write title and text in English; put the user's own words, when not English, in original. Paths, user and host names, e-mails and URL credentials are rewritten; a suspected secret or a private term refuses the draft.

Then you MUST: show the user the returned title and body verbatim (no summary, no rewording) with the content hash; list the similar posts and let the user choose: file new (feedback_submit), add to one of them (draft again with comment_on), or not send. Any change means a new draft.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "kind": "<kind>"
    },
    "name": "feedback_draft"
  }
}

Input Schema

NameRequiredDescriptionDefault
bodyNoWhat happened (bug), the problem (idea), or the comment text
kindYes"bug", "idea", "security" or "question"
stepsNoSteps to reproduce (bug only, required for a new bug)
titleNoEnglish title of a new Issue or Discussion (not for comments)
authorNoYour agent name (sub-agents: the OS name given at start)
originalNoThe user's own words when not in English (shown folded)
proposalNoHow the user imagines it working (idea only)
comment_onNoEmpty for a new post; "issue:12" or "discussion:5" to comment on one
alternativesNoWorkarounds tried or tools that almost do it (idea only)
original_languageNoLanguage code of original, e.g. "zh"

Output Schema

NameRequiredDescriptionDefault

No arguments

Schema Changelog

Changes observed during successful MCP inspections.

  1. Addedv1.17.1

TDQS

A4.8/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries the full burden and does: it discloses deterministic redaction (paths, user/host names, e-mails, URL credentials), refusal behavior for suspected secrets or private terms, a mandatory memory lookup for private-term rules, and that kind=security files nothing but returns a private link. It also states a post-condition obligation (show title and body verbatim with content hash, present similar posts, any change requires a new draft).

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Purpose and scope are front-loaded and nearly every sentence encodes a rule, so length is largely earned. It is dense and run-on in the middle (kind taxonomy, redaction, and the MUST block run together), which makes it slower to scan than it needs to be, but nothing is truly redundant.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 10-parameter drafting tool with an output schema, the definition covers prerequisites, kind selection, privacy handling, and the required follow-up workflow, including what to show the user. Return format details are rightly left to the output schema, and no agent would need to guess how to call it correctly.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so baseline is 3, but the description adds real meaning beyond it: it defines how to pick 'kind' from user language ('broken' vs 'I want', ask if ambiguous, two drafts if both), clarifies that comment_on means drafting again to attach to an existing issue/discussion, and that title/body must be English while 'original' holds the user's own words.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

Opens with a specific verb+resource ('Draft feedback to the AI Team OS maintainers') and immediately splits the outputs into a GitHub Issue vs. a Discussions idea. It is clearly separable from sibling feedback_submit, which the description names as the filing step rather than the drafting step.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Explicit when-to-use ('only after the user said they want to send feedback', may be offered on an OS error) and when-not ('only for AI Team OS itself…; feedback about Claude Code, Codex or a model goes to their own channels'). It also routes between kinds and names the alternatives: file new via feedback_submit, or draft again with comment_on.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server

Other Tools