agentleFS
June74June74/agentic-workflow/skills/decision-record/SKILL.md

decision-record

Skill0 starsChanged 3 months agoMITView on GitHub

Use after every convened decision, and at the start of every convene to read prior decisions. Maintains the append-only DECISIONS.md index that keeps the three scrum-master models a single identity.

6 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Files
SKILL.md

Decision records

The Scrum Master is three models sharing one identity. Nothing in the runtime guarantees they share a memory — each is a separate session with its own context. .agentic/DECISIONS.md is what makes them one agent instead of three agents contradicting each other in front of the teams below.

Read before you decide

At the start of every convene, read .agentic/DECISIONS.md in full. Not a summary of it. If a step would reverse a recorded decision, that alone flags the step (see flag-check), so the reversal gets recorded too and the history stays honest.

The index is append-only. Never edit or delete a past entry. A decision that turned out wrong gets a new entry that supersedes it and says so — the wrong decision and the reason it was reversed are both worth keeping.

Write after you decide

Append one line to .agentic/DECISIONS.md:

| DR-0007 | 2026-08-14 | Briefing cross-domain exposure | picked draft B | reversibility | convenes/DR-0007/ |

And write the full record to .agentic/convenes/<id>/decision.md using templates/decision-record.md. It must contain:

  • What was decided — stated so someone can act on it without reading the drafts
  • The criterion that decided it — from the pre-named criteria
  • What was rejected, and why — the most valuable and most skipped section
  • Derived acceptance criteria, if the Scrum Master inferred any from the PM's goal. Writing them here is what lets the PM veto them at a gate instead of discovering at G3 that the system graded its own homework.
  • Flag decision — was this flagged, by whom, under which trigger

Never put an example row in the live index

DECISIONS.md is read in full by every council member, and a commented-out example row is not invisible to a language model — or to grep, which does not respect HTML comments. An example that reuses a plausible ID and title reads as a real prior decision, which is exactly the divergent history this file exists to prevent. The format belongs in this skill and in templates/, never in the file itself. (This shipped once and was caught on the first dry run.)

Numbering

DR-NNNN, zero-padded, monotonic, never reused. If two convenes run concurrently, the orchestrator assigns numbers — members never self-assign.

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.