Scorecard

/scorecard -- Five Pillars Scorecard

Runs all five pillar checks (/clarity, /quality, /stability, /speed, /team) against the same project and produces one combined leadership scorecard.

When to invoke

• Monthly or quarterly project health reviews
• Before a client or executive check-in -- this is the report meant to be shared, not hidden in a terminal (Building Relationships)
• When taking over an unfamiliar or struggling project, to get inventory fast

What it does

1. Runs each of the five pillar checks in order (they build on each other -- clarity first, team last)
2. Produces a single Red / Yellow / Green scorecard across all five pillars, with a Trend column vs. the prior run when history is available
3. Calls out the one pillar that is the current bottleneck and the one action to take next -- the book's guidance is to go "an inch deep" on each pillar rather than trying to perfect one before starting another

Inputs

Same sources as the five individual skills -- see each skill's own "Inputs" section (../clarity/SKILL.md, ../quality/SKILL.md, ../stability/SKILL.md, ../speed/SKILL.md, ../team/SKILL.md). Gather what's available for each pillar before running the combined check; don't skip a pillar's data-gathering step just because it's being rolled into one report.

History / trend tracking

A single scorecard is a point-in-time snapshot. To show trend over time, read and append to a small history file in the project being assessed (not in this repo): .cm-stack/scorecard-history.jsonl. Full schema, file format, and step-by-step read/append instructions are in ../../docs/SCORECARD_HISTORY.md -- follow that doc exactly so history stays readable across skill runs and harnesses.

In short:

1. Look for .cm-stack/scorecard-history.jsonl in the project root before reporting. No file -> say so, skip the Trend column, don't invent one.
2. If found, compare this run's five scores to the most recent prior entry for the same project and add a Trend column (▲ improved, ▼ declined, ▬ unchanged, ▲▲/▼▼ for a two-step jump).
3. With 3+ prior entries, add one sentence flagging any pillar stuck at the same color across all of them.
4. After reporting, append one new JSONL line for this run (creating the file/directory if needed).

Output format

A short table (pillar, score, trend, one-line finding) followed by a single "do this next" recommendation. Written to be dropped directly into a status update or client report.

Example output

Five Pillars Scorecard -- Atlas API, Q3 check-in

Pillar	Score	Trend	Finding
Clarity	🟡 Yellow	▬	Vision agreed on informally; no written strategy below the PM.
Quality	🔴 Red	▼	3 escaped defects last 2 releases, mostly through an unreviewed hotfix path (was 🟡 Yellow last quarter).
Stability	🔴 Red	▬	No load testing on record; only 1 of 4 recent incidents got a postmortem. Red for 3 consecutive checks.
Speed	🟢 Green	▲	14 items/week, forecasting 2.5-3.5 weeks for current backlog (up from 9 items/week).
Team	🟡 Yellow	▬	Storming -- new hire under-supported, no recent retro.

Bottleneck: Quality and Stability are both red and reinforcing each other (the unreviewed hotfix path is how untested changes reach production); Stability has now been red for 3 straight checks. Do this next: require review on hotfix/* branches this week -- it's the single change that moves both pillars.

(Trend vs. Q2 check-in, read from .cm-stack/scorecard-history.jsonl; this run appended as a new entry.)

Reference

See ../../docs/FIVE_PILLARS.md and ../../docs/SCORECARD_HISTORY.md.