agentleFS
oktawianwybieralskioktawianwybieralski/coagent8/skills/coagent8-consult/SKILL.md

coagent8-consult

Skill2 starsChanged 11 days agoMITView on GitHub

Ask another local coding agent (Codex, Claude Code or Antigravity) for a second opinion through the CoAgent8 consult tool. Use when the user wants a second opinion, a check of a design or plan, help with a stubborn bug, or names CoAgent8 consult. Not for reading files, running commands or questions the workspace already answers.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 0 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Write a self-contained proposal: the question, the design, constraints and what was already tried.

    line 11
  2. Very specific:

    Put workspace paths the other agent should read itself in context_files.

    line 17
  3. Very specific:

    Leave backend, model and reasoning_effort unset unless the user named one.

    line 22
  4. Very specific:

    Before calling a model that coagent8-doctor tags ask (expensive, unpriced or on the user's ask list), tell the user its price and that the …

    line 30
  5. Very specific:

    Never call a model tagged deny: the user denied it.

    line 33

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

CoAgent8 consult

Text the user typed with this command is the question; otherwise take it from the conversation.

  1. Write a self-contained proposal: the question, the design, constraints and what was already tried. The other agent has not seen this conversation. Name files and symbols instead of pasting their content: arguments travel as escaped JSON, and the other agent reads the workspace itself. "Is the retry loop in src/a.ts (writeWithRetry) safe with concurrent writers?" is good; "thoughts?" is not.
  2. Put workspace paths the other agent should read itself in context_files; only the paths are sent. Paste only what is not in the workspace, such as an error message or a short excerpt of a log. Set task_type to architecture, debug or implementation when you want that kind of analysis; omit it for a plain question. Put required criteria in specific_questions.
  3. Leave backend, model and reasoning_effort unset unless the user named one. For a user's price limit or range (e.g. "X-Y" means output_price_below: Y), pass output_price_below (USD per million output tokens, strict) and leave model unset; the server picks the qualifying default or the most recently released qualifying model. coagent8-doctor lists models, prices and the day the price list added each model.
  4. For a follow-up on the same topic pass the session_handle from the result footer. A call without it starts a fresh agent that rereads everything.
  5. Before calling a model that coagent8-doctor tags ask (expensive, unpriced or on the user's ask list), tell the user its price and that the server will show an approval form, and let them choose it or a cheaper model. Never call a model tagged deny: the user denied it. Their answer in chat is not approval: only the form is. Never approve on the user's behalf, never change approval settings and never switch models silently to avoid the question. If the form closes without an answer, relay the refusal's ways out to the user. On a price-limit refusal, report the reason.

A turn on Codex or Claude Code has no network, web search or MCP servers; agy has its own. Paste what the other agent cannot reach, or name a backend that can: coagent8-doctor says this per backend, and a URL in an answer may not have been opened.

The answer is another model's opinion. Verify its claims against the code before acting, and tell the user where you disagree and why. The consulted agent runs read-only: it cannot edit files or run commands for you.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.