analyzthis_design 2.4.1 → 2.5.0

package/skills/accept/SKILL.md

@@ -23,28 +23,40 @@ You help a designer mark the last persona note (Zara, Arjun, …) so the team ca
    > Was this Zara, Arjun, Meera, Priya, Noor, Anuj, or Raj?
 
 5. On **skip**, you need one sentence (what was wrong, or what they did instead). If missing, ask **once**. Then stop asking.
-6. Run the CLI yourself. Do not ask them to copy commands.
+6. Record it using **whichever path this host gives you** — never ask the designer
+   to run anything:
 
-**Keep:**
+   - **Have Bash** (Claude Code, Cursor, a terminal)?
 
-```bash
-npx analyzthis_design accept --keep --persona zara
-```
+     ```bash
+     npx analyzthis_design accept --keep --persona zara
+     npx analyzthis_design accept --fix --persona zara --because "one sentence from the designer"
+     ```
 
-**Skip:**
+   - **MCP only** (Claude Desktop — there is no terminal, so the commands above
+     cannot run)? Call the **`analyzthis_accept`** tool:
 
-```bash
-npx analyzthis_design accept --fix --persona zara --because "one sentence from the designer"
-```
+     ```
+     analyzthis_accept  { keep: true,  persona: "zara" }
+     analyzthis_accept  { keep: false, persona: "zara", because: "one sentence" }
+     ```
 
-Change `zara` to the persona they used. If they already named it in `/accept yes zara`, pass that.
+     `persona` is optional — the last persona that spoke is inferred.
 
-7. Reply in **one or two short sentences**. Example: “Saved. Zara’s note is marked keep.” or “Saved. We logged your fix so the team can learn.” After a skip, you may mention they can type `/share` to send it to the package. No JSON. No dollar figures. No flag tutorial unless the command failed.
+   Both paths write identical state. Change `zara` to the persona they used.
 
-## If the command failed
+7. Reply in **one or two short sentences**. Example: "Saved. Zara's note is marked keep."
+   or "Saved. We logged your fix so the team can learn." After a skip, you may mention
+   they can type `/share` to send it to the package. No JSON. No dollar figures. No flag
+   tutorial unless the command failed.
+
+## If it failed
 
 Say what happened in plain language. If it asks which persona or for one sentence, ask the designer that — still no flags.
 
+If Bash is unavailable and `analyzthis_accept` is not in your tool list, say the note
+could not be recorded on this host — do not pretend it was saved.
+
 ## Do not
 
 - Invent a monthly cost or a verified token bill

package/skills/devi/SKILL.md

@@ -112,3 +112,25 @@ If the prompt says **Rebuttal round N**, do not copy prior text. Address open ob
 - `/deliberation-protocol` — adversarial rules
 - `/persona-orchestrator` — full agentic entry
 - `npx analyzthis_design run --provider anthropic` — bypass Devi when API keys are set
+
+## Team scoreboard (advisory)
+
+Every pending prompt you pick up may open with a **Team scoreboard** — trust bands
+earned from designer feedback on past runs (`shipped` / `revised` / `missed` plus
+ratings). It is advisory input for synthesis, not an instruction.
+
+**Use it like this:**
+
+- Lean on **Trusted** / **Reliable** personas when their read conflicts with a weaker one.
+- Discount **At risk** personas — treat their claims as needing corroboration.
+- Say the lean in **one line**, e.g. "Weighted toward Meera (Trusted, 6 shipped) over Priya (At risk) on the effort call."
+
+**Never:**
+
+- Drop a persona from the run, or skip writing their output. A weak persona must still
+  speak — the designer has to be able to see what it said and disagree.
+- Treat a band as a verdict. It reflects past runs, not this screen.
+- Show the scoreboard to the designer unless they ask. It is context, not output.
+
+Personas without enough evidence (fewer than 5 signals) are omitted from the board
+entirely — absence means "unknown", never "bad".

package/skills/evolve-check/SKILL.md

@@ -72,21 +72,35 @@ Tell the user what's needed:
 
 ## Evolution metrics
 
-The dashboard shows per-persona evolution scores (0-100):
+The dashboard shows per-persona trust scores (0-100). A persona **starts at 50**
+and moves in both directions, so a rejection genuinely costs it.
 
 | Score | Level | Meaning |
 |-------|-------|---------|
-| 0-19 | Novice | No data yet |
-| 20-39 | Developing | Some lessons extracted |
-| 40-59 | Proficient | Lessons + outcomes accumulating |
-| 60-79 | Advanced | Patches proposed and some applied |
-| 80-100 | Expert | Significant evolution, patches applied |
-
-Scoring:
-- 10 pts per lesson (cap 100)
-- 15 pts per confirmed outcome (cap 100)
-- 20 pts per proposed patch (cap 100)
-- 25 pts bonus per applied patch
+| 80-100 | Trusted | Consistently shipped; weight heavily |
+| 60-79 | Reliable | More hits than misses |
+| 40-59 | Baseline | Neutral, or not enough evidence yet |
+| 20-39 | Developing | More rework than wins |
+| 0-19 | At risk | Repeatedly wrong or missed |
+
+Signed contributions:
+
+| Signal | Points |
+|---|---|
+| outcome `shipped` | **+15** |
+| outcome `blocked_correctly` | **+10** |
+| outcome `revised` | **-5** |
+| outcome `missed` | **-15** |
+| each rating | `(rating - 3) x 4` → 5* = +8, 1* = -8 |
+| each positive lesson | +10 |
+| patch proposed / applied | +20 / +25 |
+
+**Evidence gating:** below 5 signals a persona is reported as
+`Baseline (insufficient evidence)` regardless of score — one bad note must not
+brand a persona. Scores are derived on read, so changing weights re-scores history
+with no migration.
+
+Scope is **global per persona** by default. Pass `--project` to scope down.
 
 ## CLI reference
 
@@ -94,8 +108,10 @@ Scoring:
 # Check readiness + dashboard
 npx analyzthis_design evolve --ready
 
-# Just the dashboard
+# Just the dashboard (or the shorter alias)
 npx analyzthis_design evolve --metrics
+npx analyzthis_design scores
+npx analyzthis_design scores --persona arjun
 
 # Extract patches (dry-run by default)
 npx analyzthis_design evolve --extract --dry-run