kushi-agents 3.4.1

package/plugin/reference-packs/fde/report-doctrine.md

@@ -0,0 +1,189 @@
+# FDE Report Doctrine
+
+> The hard rules every Kushi FDE report (`fde-intake`, `fde-report`, `fde-triage`) must obey when grounding claims against project Evidence.
+
+Cite as: `[source: reference-packs/fde/report-doctrine.md · packaged]`.
+
+---
+
+## Rule 1 — Source Resolution Gate
+
+Before drafting ANY FDE report, the skill MUST attempt the applicable retrieval paths and record what was retrieved vs what was attempted-and-empty vs what was attempted-and-failed. Reports that skip the gate cannot claim a complete view.
+
+Applicable retrieval paths, in order:
+
+1. **`State/` files** (if `full` profile and State/ exists) — `00_overview.md` → `09_open-questions.md`.
+2. **`Evidence/_Consolidated/` latest week** — if it exists.
+3. **`Evidence/<alias>/crm/snapshot/<record>.md`** — for CRM stage/status, MACC, ECIF, ownership.
+4. **`Evidence/<alias>/ado/snapshot/items/`** — for engagement record + work items.
+5. **`Evidence/<alias>/meetings/stream/<latest-week>.md`** — for the freshest customer-facing decisions.
+6. **`Evidence/<alias>/teams/stream/<latest-week>.md`** — for asynchronous decisions / asks.
+7. **`Evidence/<alias>/onenote/snapshot/pages/*.md`** — for durable artifacts (architecture, plans).
+8. **`Evidence/<alias>/sharepoint/snapshot/files/*.md`** — for proposals, SOWs, business cases.
+9. **`Evidence/<alias>/email/stream/<latest-week>.md`** — for sponsor / ATU correspondence.
+
+Per-alias paths repeat for every contributor in `Evidence/contributors.yml`. The skill SHOULD prefer `_Consolidated/` over per-alias once it exists.
+
+**Record per-source coverage** in the report's footer (a "Source Coverage" section). Use one of: `retrieved` / `attempted-empty` / `attempted-failed` / `not-applicable` / `not-attempted`. A report missing this section is a defect.
+
+---
+
+## Rule 2 — Recency Precedence
+
+When two sources disagree, **newer wins** — but never silently. The older signal MUST be marked **superseded** with a citation to both.
+
+Recency order (most authoritative → least):
+
+1. **Customer-facing transcript** (meetings/stream) — what was actually said.
+2. **Async customer-visible decision** (teams/stream or email/stream) — what was confirmed in writing.
+3. **CRM Confirmed Facts** — fields the FDE / EM / ATU has explicitly written-back AFTER the customer agreed.
+4. **Local durable artifacts** — OneNote pages, SharePoint proposals (signed/approved).
+5. **External links** — public references, partner sites.
+6. **Historical CRM notes** older than the freshest customer-facing source.
+
+**Do not average old and new** — pick the newest credible source, mark older as superseded.
+
+Format:
+
+```markdown
+The stage is **3 — Triage Approved** [source: ushak/crm/snapshot/AGCO-eng.md · 2026-05-08].
+
+> Older signal superseded: CRM note from 2026-04-12 said "Stage 2 — In Triage" — superseded by [source: ushak/meetings/stream/2026-05-04_meetings-stream.md · 2026-05-06 triage call decision].
+```
+
+---
+
+## Rule 3 — CRM Field Values vs Confirmed Facts
+
+CRM field updates are **internal decisions**, not confirmed facts. The FDE intake plan / status / stage fields can be edited by anyone with CRM write access **without the customer ever being asked**.
+
+Therefore: **never report a CRM field value as a confirmed fact** unless cross-referenced against the freshest customer-facing source (transcript, email, signed artifact).
+
+States like:
+
+- `Not Yet Requested`
+- `Pending`
+- `TBD`
+- `In Review`
+- `Draft`
+
+…MUST NOT be reported as finalized. Surface them with the qualifier "CRM field value — not yet customer-confirmed".
+
+When CRM disagrees with the latest customer-facing source, the report MUST call out the mismatch in `## Open Questions` AND embed a Rule 3 validation warning at the inline assertion (see Rule 4).
+
+Format:
+
+```markdown
+MACC status: **Pending** (CRM field value — not yet customer-confirmed)
+[source: ushak/crm/snapshot/AGCO-eng.md field=macc_status · 2026-05-08]
+
+> ⚠️ VALIDATION WARNING — Rule 3: CRM field "macc_status" is "Pending" but the latest sponsor email [source: ushak/email/stream/2026-05-11_email-stream.md] says "MACC is locked at $4.2M, signed last Friday". Cross-check needed; the report cannot finalize MACC until the discrepancy is reconciled.
+```
+
+---
+
+## Rule 4 — Validation Warnings Protocol
+
+For every unresolved evidence gap, missing role, missing owner, missing funding detail, blocked source, unresolved decision, or CRM-vs-customer mismatch that **materially affects interpretation** — embed a warning inline AND proceed. Do not block. Do not omit the warning.
+
+Inline format (blockquote, MUST be machine-parsable):
+
+```markdown
+> ⚠️ VALIDATION WARNING — Rule X.Y: <one-line description>
+```
+
+The `Rule X.Y` reference MUST match one of:
+
+- `Rule 1.1` — Source resolution gate skipped for a required path.
+- `Rule 1.2` — Source attempted-failed for a path material to a claim.
+- `Rule 2.1` — Recency precedence violated (newer source ignored, older quoted).
+- `Rule 3.1` — CRM field value reported as confirmed fact without customer cross-check.
+- `Rule 3.2` — CRM disagrees with latest customer-facing source.
+- `Rule 4.1` — Required role (ATU rep, Industry rep) missing from stakeholder map.
+- `Rule 4.2` — Required commercial fact (MACC, ECIF, funding model) missing.
+- `Rule 4.3` — Required outcome metric (KPI, baseline, target) missing.
+- `Rule 5.1` — Stage claim not supported by stage gate / exit criteria in `core-fde-reference.md`.
+- `Rule 5.2` — FDE fit claim not grounded against the 10-row fitness checklist.
+- `Rule 6.1` — Customer engagement legal posture (CWAA / EMA / contract status) absent.
+- `Rule 9.1` — Other unresolved gap (free-text description required).
+
+When the report is part of a triage bundle, every inline warning MUST also be mirrored to `07-validation-warnings-checklist.md` with status `open` / `resolved` / `not-applicable`.
+
+---
+
+## Rule 5 — Stage Claims Must Be Grounded
+
+Do not declare a project's FDE stage unless the Evidence supports the **exit criteria for the prior stage** as defined in `core-fde-reference.md § FDE Intake Stages, Gates, and Checklists`.
+
+If CRM says Stage 4 but Evidence only supports Stage 3 exit criteria, the report MUST:
+
+- State the CRM field value.
+- State the Evidence-supported stage.
+- Embed a `Rule 5.1` warning.
+- Surface the mismatch in `## Open Questions`.
+
+---
+
+## Rule 6 — Engagement Legal Posture
+
+When customer delivery requires a legal working agreement (e.g. CWAA — Customer Workspace Access Agreement; EMA — Enterprise Master Agreement; or a specific SOW/MSA), the report MUST explicitly assess whether the agreement is already in place AND documented in current Evidence.
+
+If absent or unclear → embed `Rule 6.1` warning AND surface in:
+
+- `fde-fitness` report → "Commercial/legal gating" row.
+- `risk-analysis` report → high-severity risk.
+- `readiness-checklist` report → blocker.
+- `executive-consolidated-report` → key risk callout.
+
+---
+
+## Rule 7 — Read-Only Discipline
+
+Every FDE report is **read-only against Evidence/**. The report skills MUST NOT:
+
+- Trigger `pull-*` skills.
+- Call WorkIQ for fresh data.
+- Make Graph / CRM / ADO writes.
+- Send any outbound message (mail, Teams DM, RSVP comment, CRM note, ADO comment).
+
+If the freshest Evidence is older than `chat.freshness_warn_days` (default 14), the report MUST:
+
+1. Warn the user at the top of the report (recommended phrasing: "Freshest evidence is N days old — consider `@Kushi refresh <project>` before relying on this report for high-stakes decisions").
+2. Continue to generate the report from what's on disk.
+3. Never silently auto-refresh.
+
+---
+
+## Rule 8 — Citation Density
+
+Every assertion (every fact, name, date, dollar figure, decision, stage claim, FDE-fit signal, risk) MUST carry a citation. Citations point to one of:
+
+- `[source: <alias>/<source>/snapshot/<file>.md as of YYYY-MM-DD]`
+- `[source: <alias>/<source>/stream/YYYY-MM-DD_<source>-stream.md · <event-id-or-timestamp>]`
+- `[source: _Consolidated/YYYY-MM-DD_consolidated.md]`
+- `[source: State/<NN>_<name>.md]` (full profile only)
+- `[source: reference-packs/fde/<file>.md · packaged|user-override|project-override]`
+
+A paragraph without at least one inline `[source: …]` is a defect. The FDE report skills MUST surface this defect as a `Rule 8.1` validation warning if any paragraph lacks citation at write time.
+
+---
+
+## Rule 9 — Privacy Posture (External-Facing Reports)
+
+For `customer` shape (or any report intended for an external audience):
+
+- Suppress internal MACC/ECIF dollar figures unless previously shared with the customer (per CRM `commercial_shared_externally` field or explicit user confirmation).
+- Suppress MS staffing detail (FDE crew composition, allocation %, cost rates).
+- Suppress FDE-internal stage terminology ("Stage 3 — Triage Approved" → "currently in triage").
+- Suppress CRM record IDs and internal ADO work item IDs.
+- Suppress attendee emails (use display names only).
+
+Internal reports (`weekly`, `short`, `long`, `handoff`, full triage bundle) include all of the above.
+
+---
+
+## See also
+
+- `core-fde-reference.md` — the operating model these rules ground against (FDE definition, anti-patterns, fitness checklist, stage gates, CRM status meanings, risk categories).
+- `intake-questions.md` — the 6 FDE Intake questions + role rules consumed by `fde-intake`.
+- `README.md` — pack overview and 3-layer override.

package/plugin/skills/aggregate-project/SKILL.md

@@ -0,0 +1,72 @@
+---
+name: "aggregate-project"
+version: "1.0.0"
+description: "Pull-only multi-source aggregation for a project. Runs every enabled pull-* skill + consolidate-evidence, but DOES NOT run build-state. Produces a complete Evidence/ folder (the public contract) without Kushi's State/ overlay — intended for standalone use and for integration with external rollup systems."
+---
+
+# Skill: aggregate-project
+
+## Purpose
+
+`aggregate-project` is Kushi's **pull-only** producer. It captures evidence from every enabled source and merges per-user streams, but stops short of building State/. Use it when:
+
+1. You want raw, multi-source evidence on disk and will render reports yourself (external rollup repo, BI tooling, custom templates).
+2. You want to refresh evidence frequently but rebuild State/ only on a slower cadence (call `@Kushi state <project>` separately when needed).
+3. You're installed at the `core` profile, where `build-state` doesn't exist.
+
+This skill **never** writes to `State/`. It only writes to `Evidence/`.
+
+## Sequence
+
+1. **Resolve project** — fuzzy-match against `m365-mutable.json knownSections` -> `active_projects` -> `<engagement-root>` subfolders.
+2. **Resolve window** — default = since last `aggregate`/`refresh`/`bootstrap` watermark in `Evidence/run-log.yml`; fallback 7 days if no watermark.
+3. **Pull each enabled source in turn** (`pull-email`, `pull-teams`, `pull-meetings`, `pull-onenote`, `pull-sharepoint`, `pull-crm`, `pull-ado`) — each writes its own snapshot/ + stream/ files under `Evidence/<alias>/<source>/`.
+4. **Run `consolidate-evidence`** — merges all aliases' streams into `Evidence/_Consolidated/<YYYY-MM-DD>_consolidated.md`. Skip silently if only one alias exists.
+5. **Update `Evidence/run-log.yml`** — record this aggregate run, its watermark, the sources pulled, the alias that produced it.
+6. **DO NOT run `build-state`.** State/ is not the concern of this skill.
+
+## Output contract
+
+After a successful run, the following exists on disk under `<engagement-root>/<project>/`:
+
+```
+Evidence/
+├── run-log.yml                                              <- manifest (public)
+├── <alias>/
+│   ├── email/stream/YYYY-MM-DD_email-stream.md              <- stream shape
+│   ├── teams/{snapshot,stream}/...                          <- snapshot + stream
+│   ├── meetings/{snapshot,stream}/...
+│   ├── onenote/{snapshot,stream}/...
+│   ├── sharepoint/{snapshot,stream}/...
+│   ├── crm/{snapshot,stream}/...                            <- if enabled
+│   └── ado/{snapshot,stream}/...                            <- if enabled
+└── _Consolidated/
+    └── YYYY-MM-DD_consolidated.md                            <- if multi-user
+```
+
+`State/` is **NOT** touched. Existing `State/*.md` files (from earlier `refresh`/`state` runs) remain unchanged.
+
+See `docs/reference/evidence-contract.md` for the full schema (filename rules, frontmatter, citation format).
+
+## What this skill never does
+
+- Never runs `build-state`. If the user wants State/, they say `@Kushi state <project>` or `@Kushi refresh <project>` separately.
+- Never sends outbound messages.
+- Never reads/writes outside the resolved project's folder.
+
+## How users invoke this
+
+- `@Kushi aggregate <project>` (default — since last watermark)
+- `@Kushi aggregate <project> last N days`
+- `@Kushi aggregate <project> since <date>`
+- `@Kushi aggregate <project> <from>..<to>`
+
+## Relationship to other verbs
+
+| Want to... | Use |
+|---|---|
+| Just capture evidence | `aggregate` |
+| Capture + rebuild State/ | `refresh` (= aggregate + build-state) |
+| Rebuild State/ only, no new pulls | `state` |
+| Capture from one source only | `pull <source>` |
+| Merge per-user streams without pulling | `consolidate` |

package/plugin/skills/ask-project/SKILL.md

@@ -0,0 +1,162 @@
+---
+name: "ask-project"
+version: "2.0.0"
+description: "Read-only Q&A over an already-bootstrapped project. Answers natural-language questions about a project (status, decisions, MACC, stakeholders, risks, action items, timeline, specific docs/meetings) using only the project's State/, Evidence/, and snapshot/ files. Cited; no source pulls; no outbound."
+---
+
+# Skill: ask-project
+
+Run this whenever the user asks a question about a known project — even without an explicit verb.
+
+The user does NOT need to type `/ask-project` or `@Kushi ask`. If the message:
+
+1. **Names a project** (matches `<engagement-root>/<project>/Evidence/`, `TempAssets/<project>/external-context/`, or `discovery.project_aliases` in any `.settings.yml`), AND
+2. **Asks a question or requests a fact / status / summary** (interrogatives: what / who / when / where / why / how / status of / list / show / summarize / compare across the same project)
+
+…then dispatch here. **Producer verbs (bootstrap / refresh / pull / state / consolidate / status) win over `ask` if the message is unambiguously a producer verb.** When ambiguous, ask the user.
+
+## Hard rules
+
+- **Read-only.** No `pull-*`, no Graph writes, no WorkIQ calls during answering. Source pulls are a separate user-initiated step.
+- **No cross-project bleed.** Answer only from the resolved project's own folders. If the question requires comparing two projects, ask the user to confirm and run `ask-project` per project, then merge in the answer.
+- **Citation Ledger applies.** Every fact, decision, person, date, $ figure carries inline `[source: <alias>/<folder>/<file> · YYYY-MM-DD]` per `instructions/citation-ledger.instructions.md` and `instructions/answer-from-evidence.instructions.md`. If the corpus does not support an answer, say so explicitly — never invent.
+- **Privacy posture.** Same as Kushi global: no outbound messages, no leaking attendee emails into anything visible to others. Answer goes to chat only.
+- **Freshness gate, not freshness auto-fix.** If the freshest source relevant to the question is older than `chat.freshness_warn_days` (default 14), warn the user and offer `@Kushi refresh <project>` — but NEVER auto-refresh.
+- **Reference packs may be consulted for domain doctrine.** When the question maps to a known reference-pack domain (currently only `fde/` — FDE stages, fitness, CRM status meanings, intake gates, risk categories, "MACC", "is this FDE-fit"), ALSO load the matching reference pack as additional grounding using the 3-layer override order (project → user → packaged). Cite reference-pack assertions with `[source: reference-packs/<pack>/<file>.md · <layer>]` where layer is `packaged` / `user-override` / `project-override`. Reference-pack content NEVER overrides project Evidence/ for facts about *this* project; it only provides definitions, gates, and rubrics.
+
+## Inputs
+
+- `<project>` — fuzzy-matched project name. If multiple plausible matches, ask the user.
+- `<question>` — the user's natural-language question. May span topics; pick the dominant topic for primary file selection.
+
+## Steps
+
+### Step 1 — Resolve project root
+
+In order, stop on first hit:
+
+1. Exact `<engagement-root>/<project>/Evidence/` folder.
+2. Fuzzy match against `discovery.project_aliases` in any `<engagement-root>/<project>/Evidence/<alias>/.settings.yml`.
+3. `<engagement-root>/<project>/external-context/` (lighter-weight projects without full bootstrap).
+4. `~/.copilot/project-evidence.yml` `active_projects` list.
+
+If zero matches → ask user. If multiple → list candidates and ask user to pick. **Echo the resolved root path before answering** so the user can confirm.
+
+### Step 2 — Classify the question
+
+Map the question to one or more **answer kinds**, which determines which files to load:
+
+| Question shape | Primary files |
+|---|---|
+| "what is the status of …", "where are we on …" | `State/00_overview.md`, latest `Evidence/<alias>/_Consolidated/`, `Evidence/run-log.yml` |
+| "what's been decided about …", "did we decide …" | `State/01_decisions.md` |
+| "who is the EM / FDE / stakeholder for …", "who's working on …" | `State/02_stakeholders.md`, `Evidence/contributors.yml` |
+| "what's the architecture / solution / approach …" | `State/03_architecture-and-solution.md`, snapshot SharePoint files matching `Architecture/` |
+| "what happened in the meeting on …", "what was discussed at …" | `Evidence/<alias>/Meetings/<date>_*.md` matching the date/topic |
+| "what action items / next steps / open asks …" | `State/05_action-items.md`, `State/09_open-questions.md` |
+| "what risks / blockers / dependencies …" | `State/06_risks-and-issues.md` |
+| "what's the timeline / when is …", "milestone for …" | `State/07_timeline-and-milestones.md` |
+| "what's the MACC / ECIF / commercials / $ size for …" | grep `MACC`, `ECIF`, `\$[0-9]`, `consumption commitment` across `State/` + `Evidence/<alias>/_Consolidated/` |
+| "what does the deck / SOW / doc say about …" | `<project>/SharePoint/snapshot/files/*.md` matching the doc name; fallback to `tree.md` to locate |
+| "what was in the email from X about Y" | grep across `Evidence/<alias>/Email/` |
+| "what's in OneNote about …" | `Evidence/<alias>/OneNote/` + `external-context/*onenote*.md` |
+| "summary of the project" | `State/00_overview.md` (this file IS the answer) |
+
+If the question doesn't fit, default to: `State/00_overview.md` + greppable scan of all `State/*.md` for topic keywords.
+
+### Step 3 — Load files (cheapest → richest, stop when sufficient)
+
+1. Load the **primary files** identified in Step 2.
+2. If primary files don't have the answer, expand to: latest week's `Evidence/<alias>/_Consolidated/` file.
+3. If still insufficient, expand to per-source latest week files for the relevant source.
+4. If still insufficient, scan `external-context/`.
+5. If still insufficient, scan `<project>/SharePoint/snapshot/files/` (these are AI-summarized doc bodies).
+6. If still insufficient → answer **"I don't have evidence for that. Want me to refresh `<source>` for `<project>`?"**
+
+**Do not** read every file in the project — that's wasteful. Stop as soon as the answer is supported by ≥1 cited source.
+
+### Step 4 — Freshness gate
+
+Before answering, read `Evidence/run-log.yml` (or `.settings.yml` `last_runs:`). For each source actually used in the answer:
+
+- If `last_pulled` is within `chat.freshness_warn_days` (default 14) → no warning.
+- If older → emit a one-liner before the answer: `> ⚠️ Freshness: <source> last pulled <YYYY-MM-DD> (<N> days ago). Answer reflects evidence as of that date. Run @Kushi refresh <project> to update.`
+
+### Step 5 — Compose the answer
+
+Format:
+
+```
+**<project>** — <one-line restatement of the question>
+
+<answer in plain prose or short bullets, every assertion cited inline>
+
+**Sources used**
+- `<alias>/<folder>/<file>` · YYYY-MM-DD
+- `<alias>/<folder>/<file>` · YYYY-MM-DD
+
+**Confidence**: high | medium | low — <one-line reason>
+
+<optional: "Want me to refresh X to get fresher evidence?" or "I noticed open question Y in 09_open-questions.md — want me to surface it?">
+```
+
+**Confidence rubric**:
+- **high** — answer comes from `State/*.md` (already curated + cited), or from ≥2 corroborating Evidence files, AND freshness OK.
+- **medium** — answer comes from a single Evidence file, OR State file is older than the underlying Evidence change, OR freshness 14–30 days.
+- **low** — answer is partial/inferred from snapshot summaries only, OR freshness >30 days, OR conflicting evidence.
+
+### Step 6 — Surface follow-ups (no sends)
+
+If the answer touches a Pending Outbound, an Open Question, or a Risk, mention it as a one-liner suggestion. **Never offer to send anything.** Per global rule, recommendations go into `State/09_open-questions.md` only — `ask-project` does not write either, it just points.
+
+## Stop conditions
+
+- Project unresolved → ask user.
+- Question is actually a producer verb in disguise ("refresh AGCO", "pull email for X") → defer to that verb's skill.
+- Corpus does not support the answer → say so + offer refresh.
+- Evidence folder missing entirely → reply: "Project `<X>` is not bootstrapped. Run `@Kushi bootstrap <X>` first."
+
+## Triggers
+
+Auto-trigger when the user message contains a project name AND any of:
+
+- "what is …", "what's …", "what was …"
+- "who is …", "who's …", "who was …"
+- "when is …", "when did …", "when's …"
+- "status of …", "where are we on …", "where do we stand on …"
+- "show me …", "list …", "summarize …"
+- "is there …", "do we have …", "did we …", "was there …"
+- "tell me about …"
+- Bare project-name + topic (e.g. "AGCO MACC", "Noridian risks", "HCA next steps")
+
+Explicit triggers also accepted:
+
+- `@Kushi ask <project> <question>`
+- `/ask-project <project> <question>`
+- `ask kushi about <project>: <question>`
+
+## Examples
+
+### Example 1 — Status question
+> User: "what's the status of AGCO?"
+> → Resolve `Engagement Assets/AGCO/`. Load `State/00_overview.md` + `run-log.yml`. Answer with cited bullets + freshness line.
+
+### Example 2 — Specific fact
+> User: "what's the MACC for AGCO and is it confirmed?"
+> → Load `State/00_overview.md`, `State/01_decisions.md`, grep `MACC` across `_Consolidated/`. Answer with $ figure + confirmation status, citing each.
+
+### Example 3 — Doc question
+> User: "what does the FDE intake deck say about Noridian's MPXe scope?"
+> → Resolve Noridian. Read `SharePoint/snapshot/files/` matching "intake" + "MPXe". Cite slide-level if available.
+
+### Example 4 — Cross-project (refused)
+> User: "compare AGCO and HCA on MACC sizing"
+> → "I'll answer per project to keep evidence boundaries clean. First AGCO: …  Then HCA: …"
+
+### Example 5 — Stale evidence
+> User: "what's the latest on Noridian?" (last pull was 21 days ago)
+> → Emit freshness warning, then answer from what's there, then offer `@Kushi refresh Noridian`.
+
+### Example 6 — Missing evidence
+> User: "what did Mark say in last week's call about pricing?"
+> → If no Meetings file for that week: "I don't have a meetings transcript for week of YYYY-MM-DD. Want me to pull meetings for AGCO since YYYY-MM-DD?"

package/plugin/skills/bootstrap-project/SKILL.md

@@ -0,0 +1,129 @@
+---
+name: "bootstrap-project"
+version: "2.1.0"
+description: "First-time setup for a project: machine preflight (WorkIQ check, optional az login), side-by-side config (templates -> live filled), engagement-root + project resolution, initial 30d snapshot+stream pull across all enabled sources. Builds State/ only on the `full` profile (skipped on `standard`). Idempotent — safe to re-run."
+---
+
+# Skill: bootstrap-project
+
+Run this when the user says any of: "bootstrap a new project", "set up project evidence for `<X>`", "add me to project `<X>`", "I'm new to `<X>`", "do it all for `<X>`" (and the project has no Evidence/ folder yet).
+
+## Profile-aware behavior
+
+This skill is **profile-aware** as of Kushi v3.3.0:
+
+- On `standard` profile (the default) — Step 6 (build-state) is **skipped**. The project's `State/` folder is NOT scaffolded; Evidence/ is the public contract.
+- On `full` profile — Step 6 runs and `State/` is scaffolded + built.
+
+The active profile is read from `kushi-install.json#profile` next to this agent file. If absent, default to `standard`.
+
+## Inputs
+
+- `<project>` — engagement name (fuzzy-matched per `engagement-root-resolution.instructions.md`).
+- `<window>` — defaults to **last 30 days** (override with `last N days` / `since <date>`).
+- Implicit: current contributor `<alias>` (from personal config).
+
+## Steps
+
+### Step 1 — Machine preflight (SETUP)
+
+Verify in order. Stop on hard failures.
+
+1. **OS + host runtime** — display OS + Node/PowerShell version. Informational.
+2. **WorkIQ install (REQUIRED, hard stop)** — read `<USER_HOME>/.copilot/project-evidence.yml workiq.cli_path`. If missing, probe known paths:
+   - `$HOME\.copilot\bin\workiq.cmd`
+   - `$env:LOCALAPPDATA\Programs\WorkIQ\workiq.cmd`
+   - `$env:ProgramFiles\WorkIQ\workiq.cmd`
+   If found, persist path. If not found, ask user for path (or to install). Test with `<workiq.cli_path> --help`. Without WorkIQ, M365 sources will all fail — STOP.
+3. **Conditional az login** — only if `<engagement-root>/.project-evidence/crm/config.yml` OR `<engagement-root>/.project-evidence/ado/config.yml` exists. Per `az-auth-conditional.instructions.md`. Soft warning on failure, never blocking.
+4. **Engagement-root resolution** — per `engagement-root-resolution.instructions.md`. Persist to `<USER_HOME>/.copilot/project-evidence.yml engagement_root` if newly resolved.
+
+Display SETUP summary table with ✅ / ⚙️ / ❌ / ⚠️ / ➖ markers.
+
+### Step 2 — Side-by-side config bootstrap
+
+Per `side-by-side-config.instructions.md`. For each missing live file, copy the template into place and ask the user ONLY for blank required fields. Group questions logically.
+
+Required live files:
+
+| Live file | Template source |
+|---|---|
+| `<USER_HOME>/.copilot/project-evidence.yml` | `templates/init/project-evidence.template.yml` |
+| `<engagement-root>/.project-evidence/m365/m365-auth.json` | `templates/init/m365-auth.template.json` |
+| `<engagement-root>/.project-evidence/m365/m365-mutable.json` | `templates/init/m365-mutable.template.json` |
+| `<engagement-root>/<project>/integrations.yml` | `templates/init/project-integrations.template.yml` |
+| `<engagement-root>/<project>/Evidence/contributors.yml` | `templates/init/project-contributors.template.yml` |
+| `<engagement-root>/<project>/Evidence/<alias>/.settings.yml` | `templates/init/project-user-settings.template.yml` |
+
+Optional (only if user enables CRM/ADO):
+
+| Live file | Template source |
+|---|---|
+| `<engagement-root>/.project-evidence/crm/config.yml` | `templates/init/crm-config.template.yml` |
+| `<engagement-root>/.project-evidence/ado/config.yml` | `templates/init/ado-config.template.yml` |
+
+### Step 3 — Project folder scaffold
+
+Create the project folder structure (per `engagement-root-resolution.instructions.md`):
+
+```
+<engagement-root>/<project>/
+   integrations.yml
+   Evidence/
+      run-log.yml                ← initial: empty watermarks
+      contributors.yml
+      <alias>/.settings.yml
+   State/                        ← FULL PROFILE ONLY
+      00_overview.md   01_decisions.md   02_stakeholders.md
+      03_architecture-and-solution.md   04_workshops-and-key-meetings.md
+      05_action-items.md   06_risks-and-issues.md
+      07_timeline-and-milestones.md   08_artifacts-and-deliverables.md
+      09_open-questions.md
+```
+
+The `State/` subtree is created **only on `full` profile**. On `standard`, only `Evidence/` is scaffolded. State files (when scaffolded) come from `templates/state/*.template.md`.
+
+### Step 4 — Initial pull (last 30 days)
+
+Dispatch to each enabled per-source skill with `--window last 30 days`:
+
+1. `pull-email`
+2. `pull-teams`
+3. `pull-meetings`
+4. `pull-onenote`
+5. `pull-sharepoint`
+6. `pull-crm` (if enabled)
+7. `pull-ado` (if enabled)
+
+Each produces snapshot/ + stream/ output per `snapshot-vs-stream.instructions.md`.
+
+### Step 5 — Consolidate (single contributor = no-op)
+
+If multiple contributors already exist, dispatch `consolidate-evidence`. For first-time bootstrap (one user), skip.
+
+### Step 6 — Build state (FULL PROFILE ONLY)
+
+If `kushi-install.json#profile` is `full` → dispatch `build-state` to render `State/*.md` from the freshly-pulled Evidence.
+
+If `standard` (or `core`) → **skip this step**. Note in the run summary: *"State/ rollup skipped (profile = `standard`). To enable, re-install with `npx github:gim-home/kushi --clawpilot --profile full --force`."*
+
+### Step 7 — Verify + summary
+
+Per `side-by-side-config.instructions.md` "Verification" rule, list all live config files (path + size + last-write time). Show:
+- SETUP summary table
+- Side-by-side config table (templates ↔ live files)
+- Per-source pull table (snapshot count + stream weeks covered)
+- New Open Questions count
+
+End with a one-liner pointing at Q&A:
+
+> ✅ `<project>` is bootstrapped. You can now ask questions naturally — e.g. *"what's the status of `<project>`?"*, *"who is the EM on `<project>`?"*, *"what's the MACC for `<project>`?"*. No `@Kushi ask` prefix needed; the `ask-project` skill auto-dispatches when a project name + question is detected.
+
+## Triggers
+
+- "bootstrap a new project"
+- "set up project evidence for `<name>`"
+- "add me to project `<name>`"
+- "I'm new to `<name>`, set me up"
+- "do it all for `<name>`" (when project has no Evidence/ folder yet)
+- "redo onboarding for `<name>`" (forces re-prompt of all config questions)

package/plugin/skills/build-state/SKILL.md

@@ -0,0 +1,69 @@
+---
+name: "build-state"
+version: "2.0.0"
+description: "Render <project>/State/*.md from existing Evidence (snapshot/ + stream/). No source pulls — pure re-render. Latest fact wins on conflicts. Every assertion cited. Always updates Open Questions."
+---
+
+# Skill: build-state
+
+Run this when the user says: "regenerate state for `<X>`", "rebuild State", "@Kushi state `<X>`". This skill makes NO source calls — it only reads Evidence and writes State.
+
+## Inputs
+
+- `<project>` — already-bootstrapped project.
+
+## Steps
+
+### Step 1 — Read all Evidence
+
+For the resolved `<project>`:
+- Walk `Evidence/<alias>/<source>/snapshot/**/*.md` (current truth per entity).
+- Walk `Evidence/<alias>/<source>/stream/**/*.md` (chronological events).
+- Walk `Evidence/_Consolidated/**/*.md` (multi-user merged where present).
+
+### Step 2 — For each State file (00–08), regenerate
+
+| State file | Pulls primarily from |
+|---|---|
+| 00_overview.md | snapshot/CRM/<id>.md, snapshot/sharepoint/files/*overview*.md, snapshot/onenote/pages/*overview*.md |
+| 01_decisions.md | stream/meetings/* (decision lines), stream/teams/* (decision lines), stream/email/* (decision lines), snapshot/ado/items/* (state changes) |
+| 02_stakeholders.md | snapshot/teams/chat-roster.md, snapshot/crm/<id>.md (engagement contacts), stream/email/* (sender/recipient analysis) |
+| 03_architecture-and-solution.md | snapshot/onenote/pages/*architecture*.md, snapshot/sharepoint/files/*arch*.md, stream/meetings/* (arch discussions) |
+| 04_workshops-and-key-meetings.md | snapshot/meetings/series-index.md, stream/meetings/* (per-meeting blocks) |
+| 05_action-items.md | stream/meetings/* (action lines), stream/teams/* (action lines), snapshot/ado/items/* |
+| 06_risks-and-issues.md | stream/meetings/* (risk lines), stream/email/* (escalations), snapshot/ado/items/* (risk-tagged) |
+| 07_timeline-and-milestones.md | snapshot/crm/<id>.md (milestones), snapshot/ado/items/* (iterations), stream/meetings/* (date commits) |
+| 08_artifacts-and-deliverables.md | snapshot/sharepoint/tree.md + files/*, snapshot/onenote/pages/* |
+
+### Step 3 — Conflict resolution
+
+- Latest fact wins on conflicts (compare `as of YYYY-MM-DD` snapshots and stream event timestamps).
+- Preserve superseded values via `Supersedes:` line under the new entry.
+- Each State file ends with a `## Conflicts` section listing every superseded value with both citations.
+
+### Step 4 — Citation density
+
+Per `citation-ledger.instructions.md`: every assertion in every State file MUST have an inline citation. Snapshot citations use `as of YYYY-MM-DD`; stream citations use `· YYYY-MM-DD`. Where a fact is BOTH "currently true" AND "decided on date X", cite both.
+
+### Step 5 — Open Questions (mandatory every run)
+
+Update `<project>/State/09_open-questions.md`:
+- Scan new evidence for items that are conflicting, second-hand, low-fidelity, or that prompt a follow-up question → add as new `Q-NNN` rows under 🔴 Open / 🟡 Partial. Set Fidelity (Low/Medium/High), Category, Source, Asked Of, Needed By.
+- For every existing 🔴/🟡 row, check if new evidence answers it. If yes: flip to 🟢 Resolved, fill Answer + Source of Answer + Resolved On, fold the answer into the appropriate State file (00–08), record path in **Folded Into**. Never delete rows.
+- If a question is no longer relevant (scope cut, decision overtaken), flip to ⚫ Stale with reason.
+- Bump the file's `As of:` line. Briefly mention in run-log how many opened / partial / resolved / staled.
+
+### Step 6 — Banner
+
+Top of every State file:
+
+```markdown
+> _As of: YYYY-MM-DD HH:mm. Regenerated by @Kushi build-state. Source: Evidence/ snapshots + streams._
+```
+
+## Triggers
+
+- "regenerate state for `<X>`"
+- "rebuild state"
+- "@Kushi state `<X>`"
+- "re-render State after I fixed `<file>`"