ledger-report

ledger-report

The map of what this project's decisions rest on. Report at moments the user already stops at: a commit, a pull request, or an explicit request. Never mid-task.

for p in ~/.claude/skills/decision-ledger/bin/ledger \
         ~/.claude/skills/ledger-report/bin/ledger \
         .claude/skills/ledger-report/bin/ledger \
         .agents/skills/ledger-report/bin/ledger; do
  [ -x "$p" ] && LEDGER="$p" && break
done

$LEDGER report      # the map
$LEDGER askable     # concepts due for a revisit here
$LEDGER personal    # what the user has defended across projects

The list covers both ways this can be installed: install.sh, which lays down the whole repo, and a skills registry, which copies this directory alone.

Present a map, not a curve

A rising line invites optimisation and becomes exactly the engagement metric this project exists to reject. A map answers the only useful question: where am I thin?

Mastery map: 24 decisions tracked, 21 days

  auth/          ██████░░░░  2 of 5 defended     <- thinnest area
  data-model/    █████████░  4 of 5, 1 decayed
  ui/            ██████████  7 of 7
  jobs/          ████░░░░░░  2 of 5, 1 unknown

  Transfers observed: 2 (session-storage, cache-invalidation)

Add at most one factual, non-judgemental line:

Since last week, auth/ grew by 340 lines while two of its decisions remain undefended.

Then stop. No advice, no nudging, no offer of a lesson, no encouragement. The information is the product, and anything appended to it reads as a verdict on the person.

Language

Every sentence describes the project, never the developer.

• Say: "auth/ has two decisions with nothing on record."
• Never: "you don't understand auth."

The distinction is not cosmetic. It decides whether this is a map someone consults or a rating they resent, and a rating ends up in a performance review, which is the failure mode the whole design avoids. There is deliberately no global score for the same reason: aggregating across areas produces a number that rates a person.