agentleFS
uprinteruprinter/ai-agents-team/.claude/skills/decision-protocol/SKILL.md

decision-protocol

Skill1 starsChanged 21 days agoMITView on GitHub

How to decide and how to ask: decide small reversible matters yourself and record them; ask the stakeholder only for decisions with lasting visible consequences, in at most eight lines, recommended option first. Preloaded into every team agent.

The file

Files
SKILL.md

Decisions — decide the small, ask the big briefly

Decide yourself and record it when one MR can reverse the choice and nothing outside the repository remembers it: wording, naming, formatting, styling, test data, which of two equivalent tools, how to record a deviation, accepting a check deviation that changed no shipped artifact and no persisted state. Write the choice and its reason in the artifact that already exists for it (tasks.md row, MR note, verify-log bullet, DEC-n line). Never put it to the stakeholder.

Ask the stakeholder only for a lasting, visible consequence: a contract or data shape another artifact consumes, persisted data, infra topology or project settings, auth, secrets, money, legal surface, anything irreversible, content going live under the stakeholder's name, scope changes, or a choice the stakeholder explicitly reserved.

The ask is at most eight lines: Decision: one line, Why it matters: one line, up to three numbered options with one-line consequences, the recommended option first and marked, then Reply with a number. Details: <one pointer>. No job IDs, SHAs, timestamps, paths, rule or AC IDs in the ask; they live at the pointer. One decision per message; ask the blocking one first. A relay forwards the block as is or shorter, never longer.

Found work follows the same split (decision-protocol.md §1a): trivial and inside the files you change — fix it in the same MR and record it, do not report it; a real problem — one Separate task proposed: line at close, three at most; never a mid-task offer or a list of small improvements (→ RP-52).

Before you send any ask, read ${CLAUDE_PLUGIN_ROOT}/.claude/process/decision-protocol.md — it carries the full tier table, the format, and the gates (sender self-check, coordinator return, /verify asks= count) (→ RP-45).

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.