agentleFS
dithiothreitoldithiothreitol/ea-skills/skills/ea-board/SKILL.md

ea-board

Skill3 starsChanged 3 months agoMITView on GitHub

Prepare the Architecture Board agenda and minutes from the governance log and health reports. Use when asked to prepare, run or minute an architecture board / governance review meeting, or to summarise what needs architectural decision this period.

Files
SKILL.md

Architecture Board support

The board's agenda is not invented -- it is read out of the repository. Everything that needs a human authority decision is already a finding, an expiring record, or an open assumption.

Building the agenda

Collect, in this order (the standing agenda):

  1. Expiring and expired dispensations -- easkills validate-gov (DISP003 errors, DISP006 warnings). Each is a decision: renew (why is compliance still impossible?) or close (what changed?).
  2. Open assumptions -- PROV006 items from easkills validate; each either gets confirmed into evidence, accepted for another period (say so in the minutes), or removed.
  3. Compliance assessments since last board -- governance-log/compliance/, leading with non-conformant verdicts and their follow-up state (COMP003 flags the ones going nowhere).
  4. Proposed decisions -- ADRs with status: proposed awaiting acceptance.
  5. Health movements -- easkills trend over the committed KPI snapshots names what regressed since the last one (take today's with kpi --as-of <date> --snapshot first and commit it with the minutes); easkills diff --ref <commit of the last board> lists what changed in the approved model since then, by field. Obsolescence exposure, unsupported capabilities, the review queue by owner. For a roadmap item, easkills diff --plateau <from> <to> states what the target state changes -- what enters, what leaves, and the gap that closes it.
  6. Rationalization candidates -- debt items of kind rationalization-candidate, overlapping-applications and duplicate-service. Each is a board item precisely because the tool cannot decide it: overlap is either portfolio drift or bought redundancy, and only a human knows which. The decision the board makes is one of three, and all three end in a record: consolidate (which system wins, and who funds the migration), keep both on purpose -- which means an ADR naming the reason, so the finding stops coming back every quarter -- or split the capability, because two realizers can also mean the map is drawn too coarse. Bring the realizers' TIME dispositions to the table; debt prints them next to each candidate for this reason. An overlap carried unrecorded from one board to the next is the finding, not the overlap.
  7. Service performance -- the kpi Service line: SLA breaches (REQ006) each need a disposition (fulfil, decline, renegotiate the promise), offerings nobody requests are candidates for retirement, and repeated informal asks are a missing catalog entry.
  8. Pending promotions -- staged content whose review is stuck (promote --dry-run shows what is ready).

For each item, prepare the decision to be made, not just the topic: "renew disp-onprem-legacy for 6 months or fund the migration" is an agenda item; "discuss on-premise hosting" is not.

The maturity table as a board artifact

easkills maturity --as-of <date> produces the one slide a board actually acts on — provided you bring it whole. Never reduce it to a single figure, and never present a level without the items blocking the next one: a level is a description, the blockers are the agenda. Two dimensions moving in opposite directions is the interesting case and an average would erase it.

Bring it quarterly, against your own previous run. A not measurable row is a resourcing question ("we own no reference model"), not a bad score — and it is the one row where the fix is a decision the board can take in the room.

Minutes

Minutes are the records they change, plus a summary. Every board outcome lands as a concrete edit in the same commit series: an accepted ADR (status: accepted), a renewed/closed dispensation, a confirmed assumption (evidence added, assumed removed) or its conscious re-acceptance, an updated lastReviewed. A written summary (governance-log/board-YYYY-MM-DD.md) lists attendees, decisions taken and who owns each follow-up -- but the summary points at records; it never contains governance that exists nowhere else.

Reporting back

The agenda as a numbered list of decisions-to-make with their evidence, anything that could not be prepared for lack of an owner or record, and -- after the board -- the commit(s) carrying the outcomes.

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.