eap-lean-gain

name	eap-lean-gain
description	One-shot impact scoreboard for EAP-Lean — measured values only. Counts the `eap-lean:` markers actually in this repo and aggregates any "net: -N lines" figures the review/audit skills produced this session. NEVER prints a benchmark percentage, an upstream headline number, or a per-repo cost claim: the unbuilt version was never written, so there is no baseline to subtract from. One-shot display, not a persistent mode. Use when the user says "lean gain", "what does lean save", "show lean impact", "lean scoreboard", or runs /eap-lean-gain.
license	MIT

EAP-Lean gain

Display a scoreboard when invoked. One-shot: do NOT change mode, write flag files, or persist anything.

Every number on the card must be measured in this repo, in this session. There are exactly two honest sources:

1. Debt markers — count them the same way eap-lean-debt does:

   grep -rnE '(#|//|--|;) ?eap-lean:' . \
     --exclude-dir=node_modules --exclude-dir=.git --exclude-dir=dist
   

   Each marker is one deliberate shortcut that shipped instead of the bigger build.

2. Session net estimates — any net: -N lines possible figure that eap-lean-review or eap-lean-audit emitted earlier in THIS session, and any diff you actually applied under the ladder this session (count the lines: they are in the transcript).

Scoreboard

Render plain ASCII, only the rows you have data for:

  eap-lean gain                          this repo · measured only

  Shortcuts shipped lean     <N> eap-lean: markers   (<M> missing a trigger)
  Cuts identified            net: -<N> lines possible   (review/audit, this session)
  Cuts applied               -<N> lines               (diffs applied this session)

  Next:  /eap-lean-debt   (the ledger behind the marker count)
         /eap-lean-audit  (what's still cuttable)

No markers and no session figures: say so — No measured gain yet. Run /eap-lean-review on a diff or /eap-lean-audit on the repo.

Honesty boundary

• NEVER print a savings percentage, a cost figure, or a speed multiple. Those require a benchmark A/B; run ../../bench/ with your own API key if you want them.
• NEVER print an upstream headline number — the adapted-from project's measurements are its own, not this repo's.
• NEVER claim "you saved X lines/tokens/dollars here": in a live repo the over-built version was never written, so there is no real baseline. The only real per-repo figures are counted ones — the debt ledger and session diffs — and this card shows exactly those, nothing else.

Boundaries

• One-shot display. Reads and reports only; edits nothing, changes no mode.
• "stop lean gain" or "normal mode" reverts.