ticketwright 2.0.0__tar.gz → 3.2.0__tar.gz

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/config/stack.example.asana-bq.yaml

@@ -17,6 +17,7 @@ seams:
     tool: asana
     adapter: adapters/tracker/asana.md
     transport: mcp
+    mcp: asana                          # your Asana MCP server name (adapter calls resolve as mcp__{mcp}__…)
     workspace_gid: "1122334455"
     default_project_gid: "9988776655"
     verify: null                        # MCP-only
@@ -34,6 +35,7 @@ seams:
     tool: teams
     adapter: adapters/chat/teams.md
     transport: mcp
+    mcp: teams                          # your Teams MCP server name (adapter calls resolve as mcp__{mcp}__…)
     channel: "Data Team / General"
     default_mode: draft
     always_include: [Ana]
@@ -60,7 +62,7 @@ policies:
   db_write_requires_approval: true
   chat_default_draft: true
   hyperlink_everything: true
-  commandify_everything: true
+  skillify_everything: true
   reduce_assumptions: true
   commit_plan_before_implement: true
   system_evolution: true

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/config/stack.example.azure.yaml

@@ -22,7 +22,7 @@ seams:
     org: acme-corp
     project: Platform
     done_state: Closed
-    verify: "az boards work-item show --id {default_epic}"
+    verify: "az devops project show --project {project}"
 
   warehouse:
     tool: synapse
@@ -37,6 +37,7 @@ seams:
     tool: teams
     adapter: adapters/chat/teams.md
     transport: mcp
+    mcp: teams                          # your Teams MCP server name (adapter calls resolve as mcp__{mcp}__…)
     channel: "Data Platform / General"
     default_mode: draft
     always_include: [Dana]
@@ -66,7 +67,7 @@ policies:
   db_write_requires_approval: true
   chat_default_draft: true
   hyperlink_everything: true
-  commandify_everything: true
+  skillify_everything: true
   reduce_assumptions: true
   commit_plan_before_implement: true
   system_evolution: true

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/config/stack.schema.md

@@ -33,6 +33,7 @@ policies:       # behavioral rules every skill inherits (the kit's "global rules
 | `terminal_status` | string | `Done` | The "done" workflow state (not always "Done"). |
 | `ticket_url_template` | template \| null | `https://acme.atlassian.net/browse/{id}` | How `tickets/INDEX.md` links each ticket (`{id}` token). Null/omitted → the index renders no per-ticket link. |
 | `word_limits` | map | `{tracker_comment: 100, chat: 100, pr: 200, ticket: 200}` | Hard caps the comms skills enforce. |
+| `graph_notes` | bool | `true` | Generate the Obsidian graph layer (`tickets/graph/` + `tickets/objects/`). On by default; set `false` to disable. |
 
 ## `seams`
 
@@ -57,7 +58,7 @@ and `refresh context` degrade gracefully (skip warehouse steps) when it is.
 | `db_write_requires_approval` | `true` | any skill issuing a non-SELECT — show SQL, explain, wait for `yes`. |
 | `chat_default_draft` | `true` | `chat.draft` not `chat.send` unless the user says "send it". |
 | `hyperlink_everything` | `true` | comms skills wrap every ticket-ID / file / PR in a smart link. |
-| `commandify_everything` | `true` | recurring work → `productize`, not a one-off. |
+| `skillify_everything` | `true` | recurring work → a `/productize` skill the agent can invoke, not a one-off. |
 | `reduce_assumptions` | `true` | ask before building; still document every assumption in the ticket README. |
 | `commit_plan_before_implement` | `true` | `spec-and-build` commits the spec/plan artifact before `build` (blame-free retry). |
 | `system_evolution` | `true` | `ship` retro: a failure fixes the AI layer (rule/context/command/adapter), not just the ticket. |

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/config/stack.yaml

@@ -17,6 +17,7 @@ project:
   terminal_status: Done                 # the workflow's "done" state (not always "Done")
   ticket_url_template: "https://acme.atlassian.net/browse/{id}"  # how INDEX.md links a ticket ({id} token); omit/null = no link
   word_limits: {tracker_comment: 100, chat: 100, pr: 200, ticket: 200}
+  graph_notes: true                     # Obsidian graph layer (tickets/graph + tickets/objects); false to disable
 
 seams:
   tracker:
@@ -26,7 +27,7 @@ seams:
     site: acme.atlassian.net
     cli: acli
     mcp: atlassian
-    verify: "acli jira workitem view {default_epic}"
+    verify: 'acli jira workitem search --jql "project = {key_prefix}" --limit 1'
 
   warehouse:
     tool: snowflake
@@ -69,7 +70,7 @@ policies:
   db_write_requires_approval: true
   chat_default_draft: true
   hyperlink_everything: true
-  commandify_everything: true
+  skillify_everything: true
   reduce_assumptions: true
   commit_plan_before_implement: true
   system_evolution: true

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/hooks/regenerate_ticket_index.py

@@ -46,12 +46,23 @@ def main() -> int:
     # index_data.json is source, NOT generated — editing the store SHOULD re-render.
     if changed.name in ("INDEX.md", "OBJECTS.md"):
         return 0
+    # generated graph layer (tickets/graph/, tickets/objects/) — regenerated below; never react to it
+    if any(part in ("graph", "objects") for part in changed.relative_to(tickets_dir).parts[:-1]):
+        return 0
 
-    sys.path.insert(0, str(root / "bin"))
+    # The renderer module ships with the kit (bin/), not the user's project — resolve it against
+    # CLAUDE_PLUGIN_ROOT (else this hook's own kit dir). `root` stays the project for the data.
+    kit = os.environ.get("CLAUDE_PLUGIN_ROOT")
+    bindir = (Path(kit).resolve() if kit else Path(__file__).resolve().parent.parent.parent) / "bin"
+    sys.path.insert(0, str(bindir))
     try:
-        from build_ticket_index import build_rows, render, render_objects  # type: ignore
+        from build_ticket_index import build_rows, render, render_objects, render_graph_layer, load_config  # type: ignore
         rows = build_rows(root)
+        cfg = load_config(root)
         fresh = {tickets_dir / "INDEX.md": render(rows), tickets_dir / "OBJECTS.md": render_objects(rows)}
+        graph_dirs = [tickets_dir / "graph", tickets_dir / "objects"]  # always tracked so disabling cleans up
+        if cfg.get("graph_notes", True):
+            fresh.update(render_graph_layer(rows, root))
     except SystemExit:
         return 0  # malformed index_data.json — surfaced when the agent runs the renderer
     except Exception:
@@ -61,9 +72,18 @@ def main() -> int:
         changed_any = False
         for p, txt in fresh.items():
             if (p.read_text() if p.is_file() else None) != txt:
+                p.parent.mkdir(parents=True, exist_ok=True)
                 p.write_bytes(txt.encode("utf-8")); changed_any = True
+        fresh_keys_ci = {str(p).lower() for p in fresh}  # orphan cleanup (case-insensitive for macOS)
+        for gd in graph_dirs:
+            if gd.is_dir():
+                for existing in gd.glob("*.md"):
+                    if str(existing).lower() not in fresh_keys_ci:
+                        existing.unlink(); changed_any = True
+                if not any(gd.iterdir()):
+                    gd.rmdir()
         if changed_any:
-            print("Auto-regenerated tickets/INDEX.md + OBJECTS.md (ticket folder changed). "
+            print("Auto-regenerated tickets/INDEX.md + OBJECTS.md + graph layer (ticket folder changed). "
                   "Run bin/ingest_index_records.py to refresh a ticket's curated summary.")
     except OSError:
         pass

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/hooks/session_context.py

@@ -52,10 +52,8 @@ def main() -> int:
 
     skills = sorted(p.parent.name for p in (root / ".claude/skills").glob("*/SKILL.md")) \
         if (root / ".claude/skills").is_dir() else []
-    # Hide deprecated v1 alias stubs — they route to v2 skills and shouldn't re-clutter the surface.
     commands = sorted(
         p.stem for p in (root / ".claude/commands").glob("*.md")
-        if "Deprecated" not in p.read_text(errors="replace")[:200]
     ) if (root / ".claude/commands").is_dir() else []
 
     lines = [

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/hooks/ticket_index_context.py

@@ -38,7 +38,11 @@ def discovered_total(root: Path) -> int | None:
     """True ticket count from the renderer's discovery (cheap globs, no README reads)."""
     try:
         import sys
-        sys.path.insert(0, str(root / "bin"))
+        # The renderer module ships with the kit (bin/), not the project — resolve via
+        # CLAUDE_PLUGIN_ROOT (else this hook's own kit dir); `root` stays the project for discover().
+        kit = os.environ.get("CLAUDE_PLUGIN_ROOT")
+        bindir = (Path(kit).resolve() if kit else Path(__file__).resolve().parent.parent.parent) / "bin"
+        sys.path.insert(0, str(bindir))
         from build_ticket_index import discover  # type: ignore
         return len(discover(root))
     except Exception:

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/settings.json.tmpl

@@ -1,5 +1,5 @@
 {
-  "_README": "Rendered to .claude/settings.json by /setup. Wires the kit's hooks (policy enforcement + session priming + ticket-index surfacing/auto-regen), a conservative read-only permission baseline (setup appends stack-specific query/CLI allows), and the stack statusline. The db_write_guard hook turns 'db_write_requires_approval' into a mechanical ask before any destructive warehouse statement.",
+  "_README": "Rendered to .claude/settings.json by /setup. The hooks block below is for a VENDORED (cp -r) install, where nothing else wires the kit's hooks. On a PLUGIN install (CLAUDE_PLUGIN_ROOT set), .claude-plugin/plugin.json already wires these same hooks from the plugin dir, so /setup OMITS this hooks block to avoid double-firing (double db-write prompts, double index regen) AND ADDS an 'extraKnownMarketplaces' (ticketwright github source, autoUpdate:true) + 'enabledPlugins' (ticketwright@ticketwright:true) block so the REPO opts every user into the plugin at PROJECT scope (a plugin cannot set its own install scope — the repo does; teammates are prompted to install), committed with the repo, and self-updates on formal releases only (version-string change; never un-released main commits). Do NOT add those two keys on a vendored install — there is no marketplace to enable from. permissions + statusLine are written in both modes (setup appends stack-specific CLI allows); on a plugin install setup copies statusline.sh into .claude/ so the relative command resolves. The db_write_guard hook turns 'db_write_requires_approval' into a mechanical ask before any destructive warehouse statement.",
   "hooks": {
     "PreToolUse": [
       {

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/skills/productize/SKILL.md

@@ -8,7 +8,7 @@ disable-model-invocation: true
 
 # /productize
 
-The meta-skill (`commandify_everything`). When a task recurs (quarterly sale, monthly report,
+The meta-skill (`skillify_everything`). When a task recurs (quarterly sale, monthly report,
 periodic reconciliation), this stamps out a new **folder skill** with the productized skeleton:
 a phased pipeline with hard halts, parameterized SQL/steps, a QC checkpoint, a golden-replay
 regression test, and a hard halt before any external post. `/refresh index` + the `--recurring`
@@ -25,7 +25,7 @@ Ask (AskUserQuestion) for:
   to assert; the byte-identical output file to diff against).
 
 ## Phase 2 — Stamp from the template
-1. Copy `templates/productized-skill/` → `.claude/skills/<name>/` (SKILL.md + `sql/ templates/
+1. Copy `${CLAUDE_PLUGIN_ROOT:-$CLAUDE_PROJECT_DIR}/templates/productized-skill/` → `.claude/skills/<name>/` (SKILL.md + `sql/ templates/
    bin/ golden/`).
 2. Render `SKILL.md.tmpl` with the interview answers into the canonical phase shape:
    **Phase 0** pre-flight (validate params + `verify_stack` for touched seams + an object drift

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/skills/productize/authoring.md

@@ -24,7 +24,7 @@ warehouse round-trip — cheap insurance. It asserts: zero leftover tokens · no
 balanced single-quotes / parens · flags an unquoted SQL literal. Halt on any error.
 
 ```bash
-bash bin/render_and_validate.sh sql/step.sql.tmpl asof=2026-06-30 src=T [--strict]
+bash "${CLAUDE_PLUGIN_ROOT:-$CLAUDE_PROJECT_DIR}/bin/render_and_validate.sh" sql/step.sql.tmpl asof=2026-06-30 src=T [--strict]
 ```
 
 ## Export helpers — `bin/split_and_export.sh`

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/skills/review/SKILL.md

@@ -24,7 +24,10 @@ Read-only: it reviews and re-runs, it does not edit code (the build owns fixes).
 - `= NULL` → must be `IS NULL`; missing div-by-zero guard; `SELECT *` in deliverables; functions on
   filtered columns; implicit cross-source type mismatches (missing `CAST`); hardcoded values that
   should be parameters; missing required schema/instance filters; `LEFT JOIN` predicate in `WHERE`
-  that silently becomes an inner join; `NOT IN` with nullable columns; `UNION` vs `UNION ALL`.
+  that silently becomes an inner join; `NOT IN` with nullable columns; `UNION` vs `UNION ALL`;
+  session-variable/`DECLARE` parameterization in a shipped or exported query (prefer a CTE params
+  row — portable, single-statement, export-clean; session vars pollute `--format csv` output and
+  don't survive into single-statement JDBC/ODBC clients).
 
 **② Counts & dedup** (re-run independently)
 - Re-run the row-count; compare to the documented count. **Duplicate detection is the primary test:**
@@ -42,7 +45,9 @@ Read-only: it reviews and re-runs, it does not edit code (the build owns fixes).
   maintainability). Classify each finding Critical / Should-fix / Review.
 
 **⑤ Human sign-off**
-- Output format check (CSV headers row 1, no preamble/blank rows; filenames carry record counts).
+- Output format check (CSV headers row 1, no preamble/blank rows; filenames carry record counts;
+  **ASCII punctuation only in cell values** — no em/en dashes, smart quotes, or ellipsis chars, which
+  render as mojibake in Excel/CSV viewers).
 - README completeness (assumptions enumerated, QC results, business context). Flag for the human.
 
 ## Tiers & halting

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/skills/setup/SKILL.md

@@ -46,10 +46,11 @@ include the key commented with a `# TODO` and keep going — `verify` will point
 4. Compose `.claude/config/stack.yaml` per `stack.schema.md` (chosen seams live; optional seams as
    commented blocks; the 9 policies with a one-line "when to change this" comment each). Warn if a
    chosen adapter is `status: stub`.
-5. Scaffold the repo per [scaffold.md](scaffold.md): render `AGENTS.md` (+ role focus),
-   `.claude/settings.json` (hooks + read-only CLI allows), folders, `.gitignore` (with the
-   anchored `**/final_deliverables/*.csv` export guard), the AI-layer index, and the seeded ticket
-   index.
+5. Scaffold the repo per [scaffold.md](scaffold.md): render `AGENTS.md` (+ role focus) and a one-line
+   `CLAUDE.md` (`@AGENTS.md`, so Claude Code auto-loads the rules),
+   `.claude/settings.json` (hooks — omitted on a plugin install — + read-only CLI allows), folders,
+   `.gitignore` (deliverable CSVs committed by default; PII opts out via `*.private.csv` / a
+   `private/` subfolder), the AI-layer index, and the seeded ticket index.
 
 ### Phase 4 — Verify & hand off
 6. Run `!bash "${CLAUDE_PLUGIN_ROOT:-$CLAUDE_PROJECT_DIR}/bin/selftest.sh"` (kit integrity — a failure here is fatal) and
@@ -57,3 +58,10 @@ include the key commented with a `# TODO` and keep going — `verify` will point
    **not** fatal at setup time; print its adapter's auth notes as the fix).
 7. **Report:** the chosen stack, files written, any `# TODO` keys or stub adapters, and the next
    step — `/ticket <id>` to start work, or `/setup --teammate` for a new person.
+8. **Offer to commit the scaffold.** What setup just wrote (`.claude/config/stack.yaml`, `AGENTS.md`,
+   `CLAUDE.md`, `.claude/settings.json`, `.gitignore`, `documentation/AI_LAYER_INDEX.md`, the seeded `tickets/`
+   index — plus, on a vendored install, the kit itself) is untracked; if it isn't committed, a later
+   ticket PR references rules/adapters absent from the repo's history. Offer a commit (e.g.
+   `chore: initialize ticketwright workspace`). First flag that `stack.yaml` may hold internal
+   identifiers (tracker site, warehouse project/dataset) — config, not secrets, but worth a glance
+   before committing to a public repo.

ticketwright-3.2.0/.claude/skills/setup/scaffold.md

@@ -0,0 +1,80 @@
+# Scaffold details — what `/setup` writes into a fresh repo
+
+## Global rules (`AGENTS.md`)
+Render `${CLAUDE_PLUGIN_ROOT:-$CLAUDE_PROJECT_DIR}/templates/AGENTS.md.tmpl` → `AGENTS.md` (tokens
+from `stack.yaml`: tool names, key_prefix, terminal_status, word limits, policies). Fill
+`{{role_focus}}` from `${CLAUDE_PLUGIN_ROOT:-$CLAUDE_PROJECT_DIR}/templates/roles/<role>.md` using
+`project.role` (`generalist` unless the user changed it). This is the always-loaded tier —
+keep it the rendered template; repo-specific rules get added by humans over time.
+
+Also write `CLAUDE.md` from `${CLAUDE_PLUGIN_ROOT:-$CLAUDE_PROJECT_DIR}/templates/CLAUDE.md.tmpl` — a
+one-line `@AGENTS.md` import so **Claude Code** auto-loads these rules (it reads `CLAUDE.md`; other
+agents read `AGENTS.md` directly). Keep it to that single import line.
+
+## Hooks + settings (`.claude/settings.json`)
+Render `${CLAUDE_PLUGIN_ROOT:-$CLAUDE_PROJECT_DIR}/.claude/settings.json.tmpl` → `.claude/settings.json`.
+**The `hooks` block is install-mode-dependent:**
+- **Plugin install** (`${CLAUDE_PLUGIN_ROOT}` set): `.claude-plugin/plugin.json` already wires the
+  kit's hooks from the plugin dir — **OMIT the `hooks` block** here, or they double-fire (double
+  db-write prompts, double index regen). Keep `permissions` + `statusLine`.
+- **Vendored install** (`cp -r`, no `${CLAUDE_PLUGIN_ROOT}`): **keep the `hooks` block** — nothing
+  else wires it: `db_write_guard.py` (PreToolUse) makes `db_write_requires_approval` mechanical;
+  `session_context.py` + `ticket_index_context.py` (SessionStart) prime the stack + ticket catalog;
+  `regenerate_ticket_index.py` (PostToolUse) keeps `tickets/INDEX.md` fresh on folder changes.
+
+**On a plugin install, also commit a project-scoped enablement.** A plugin can't set its own install
+scope — the *repo* opts in. Merge these two keys
+into the rendered `.claude/settings.json` so the plugin is enabled *for this repo* (committed →
+travels with the repo; teammates who open and trust it are prompted to install it; it survives the
+original author leaving) and refreshes
+itself:
+
+```json
+{
+  "extraKnownMarketplaces": {
+    "ticketwright": {
+      "source": { "source": "github", "repo": "kyle-chalmers/ticketwright" },
+      "autoUpdate": true
+    }
+  },
+  "enabledPlugins": {
+    "ticketwright@ticketwright": true
+  }
+}
+```
+
+`autoUpdate` re-installs **only when the plugin's version string changes** — i.e. only on a formal
+release (the release commit bumps `plugin.json`/`marketplace.json`/`__init__.py` in lockstep and tags
+`v*`). Between releases, ordinary commits to the default branch leave the version untouched, so
+teammates are never pulled onto un-released mid-flight work. Do **not** add these keys on a vendored
+(`cp -r`/pip) install — there's no marketplace to enable from; the kit is already in-repo.
+
+Then append the chosen warehouse/tracker/vcs **read-only** CLI allows to `permissions.allow` (e.g.
+`Bash(<warehouse_cli> …:*)`). **Statusline:** the template's `statusLine.command` is the
+project-relative `.claude/statusline.sh`, so on a plugin install **copy
+`${CLAUDE_PLUGIN_ROOT}/.claude/statusline.sh` → `.claude/statusline.sh`** so it resolves (on a
+vendored install it's already there).
+
+## Folders + `.gitignore`
+Create `tickets/{assignee_dir}/`, `documentation/`, `resources/`, `specs/` (and `ci/` if wanted).
+Render `${CLAUDE_PLUGIN_ROOT:-$CLAUDE_PROJECT_DIR}/templates/gitignore.tmpl` → `.gitignore` (merge if
+one exists). Deliverable exports (`final_deliverables/*.csv` etc.) are **committed by default** so
+results live with the ticket and show in the PR; PII/customer data opts out via a `*.private.csv`
+name or a `private/` subfolder (both gitignored). If the tracker adapter ships an attachment-download
+helper, copy it into `resources/`.
+
+## AI-layer index (`documentation/AI_LAYER_INDEX.md`)
+A one-line-each inventory of the installed skills (`setup`, `ticket`, `spec-and-build`, `review`,
+`ship`, `productize`, `refresh`), the `qc-reviewer` agent, the hooks, and the adapters in use — so
+humans and agents can find what exists.
+
+## Ticket index
+Seed an empty curated store — `tickets/index_data.json` with
+`{"schema_version": 1, "tickets": []}` — then run
+`python3 "${CLAUDE_PLUGIN_ROOT:-$CLAUDE_PROJECT_DIR}/bin/build_ticket_index.py"` to write the
+initial `tickets/INDEX.md`. From here it self-maintains (PostToolUse regen on folder changes,
+SessionStart surfacing, curated summaries at ship time). An existing backlog gets bootstrapped with
+`/refresh index --all`.
+
+That renderer also writes the Obsidian graph layer (`tickets/graph/` + `tickets/objects/`) when
+`project.graph_notes` is on (the default), committed alongside `INDEX.md`/`OBJECTS.md`.

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/skills/ship/SKILL.md

@@ -40,9 +40,13 @@ authorization, execute in order:
 7. **chat.draft** to `seams.chat.default_channel` (policy `chat_default_draft` — the human clicks
    send unless they said "send it", in which case `chat.send`). Smart links for ticket id(s),
    files, PR.
-8. **vcs.commit** (stage this ticket's paths **plus `tickets/INDEX.md` + `tickets/OBJECTS.md` +
-   `tickets/index_data.json`** — all three, or `--check` flags drift in CI; semantic message +
-   Co-Authored-By) then **vcs.open_pr** (semantic title; body = Business Impact / Deliverables /
+8. **vcs.commit** — stage this ticket's paths (deliverable files included: they're committed by
+   default so results live with the ticket and show in the PR) **plus `tickets/INDEX.md` +
+   `tickets/OBJECTS.md` + `tickets/index_data.json`** (all three, or `--check` flags drift in CI;
+   semantic message + Co-Authored-By). **Before staging, list the `final_deliverables/` files that
+   will be committed and confirm none carry PII/customer data that shouldn't be in git** — if any do,
+   have the user rename them `*.private.csv` (etc.) or move them under a `private/` subfolder (both
+   gitignored) first. Then **vcs.open_pr** (semantic title; body = Business Impact / Deliverables /
    Technical Notes / QC).
 9. **transition** the ticket toward `project.terminal_status` if appropriate.
 

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/skills/spec-and-build/SKILL.md

@@ -27,7 +27,7 @@ context-engineering core idea: AI fails from missing context, not weak models.
      `bin/recall.py --for <id>`) and reuse their SQL/QC where it fits;
    - pull the business rules from the `documentation/` glossary (the domain slice).
    Research agents return findings only; **they do not write code.**
-3. **Write the spec** from `templates/spec.md.tmpl` into the ticket's folder
+3. **Write the spec** from `${CLAUDE_PLUGIN_ROOT:-$CLAUDE_PROJECT_DIR}/templates/spec.md.tmpl` into the ticket's folder
    (`specs/<id>-<slug>.md` or `final_deliverables/`): operation type (new/alter), data grain,
    sources + join/cast rules, transformation logic, **validation gates** (the exact QC the build must
    pass), downstream impact, dev-env target (`seams.warehouse.dev_db`), and a **confidence score
@@ -41,7 +41,9 @@ context-engineering core idea: AI fails from missing context, not weak models.
 6. **Load** the committed spec (path arg or newest in the ticket's `specs/`). Treat it as the source
    of truth, but **validate each step independently** — don't blindly follow; the spec can be wrong.
 7. **Implement in small build-and-check sub-loops:** one object/step at a time. Develop against
-   `seams.warehouse.dev_db` first; parameterize values at the top; explicit `ORDER BY` on any export
+   `seams.warehouse.dev_db` first; parameterize values at the top **via a CTE params row
+   (`WITH params AS (SELECT … AS anchor) … CROSS JOIN params`), not a session `DECLARE`/`SET`** —
+   CTE params stay portable and keep CSV exports clean; explicit `ORDER BY` on any export
    (deterministic outputs).
 8. **Embed validation between steps** — after each, run the relevant gate from the spec; self-correct.
 9. **Any non-SELECT / DDL** ⇒ policy `db_write_requires_approval`: show the exact SQL, explain the

{ticketwright-2.0.0 → ticketwright-3.2.0}/.claude/skills/ticket/SKILL.md

@@ -35,9 +35,14 @@ this before?", "which tickets touched VW_X?").
 5. **Branch** via the vcs adapter, named `<id>`, off `seams.vcs.default_branch` — or a **worktree**
    when `--worktree` is passed (isolates this ticket from other in-flight work; recommended when you
    run several tickets in parallel).
-6. **Scaffold** `project.ticket_path` with `project.ticket_subdirs`; tracker adapter
-   `download_attachments` → `source_materials/` (silent if none); render
-   `templates/ticket-README.md.tmpl` → the ticket dir.
+6. **Scaffold** `project.ticket_path` with `project.ticket_subdirs` (create the subdirs empty —
+   **no `.gitkeep` placeholders**; they fill with real files during build and git picks them up
+   then); tracker adapter `download_attachments` → `source_materials/` (silent if none); render
+   `${CLAUDE_PLUGIN_ROOT:-$CLAUDE_PROJECT_DIR}/templates/ticket-README.md.tmpl` → the ticket dir.
+6b. **Refresh the catalog** so the new ticket shows up immediately — it won't otherwise, because the
+   PostToolUse index hook only fires on `Write`/`Edit` and scaffolding happens via Bash:
+   `python3 "${CLAUDE_PLUGIN_ROOT:-$CLAUDE_PROJECT_DIR}/bin/build_ticket_index.py"` (writes this
+   project's `tickets/INDEX.md` + `OBJECTS.md`; the new row shows `▱` until `/ship` curates it).
 
 ## Phase 3 — Prime context automatically (the part you never have to ask for)
 7. Follow [priming.md](priming.md), in order:

{ticketwright-2.0.0 → ticketwright-3.2.0}/CHANGELOG.md

@@ -3,6 +3,109 @@
 All notable changes to this project are documented here. Format loosely follows
 [Keep a Changelog](https://keepachangelog.com/); this project uses semantic-ish versioning.
 
+## [3.2.0] — 2026-07-06
+
+Makes a plugin install **project-scoped by default** — the *repo* commits the enablement (a plugin
+can't set its own install scope), so it travels with the repo and every teammate who opens the repo is
+prompted to install it (no marketplace to add or config to write), staying current across the team.
+
+### Added
+- **Project-scoped enablement is the default on plugin installs.** On a plugin install, `/setup` now
+  also writes an `extraKnownMarketplaces` (ticketwright github source, `autoUpdate: true`) +
+  `enabledPlugins` (`ticketwright@ticketwright`) block into the repo's committed
+  `.claude/settings.json`. It's the *repo* opting in at project scope (a plugin can't set its own install scope): teammates who
+  open and trust the repo are prompted to install it — that repo only — and it keeps working after the
+  person who set it up moves on. `autoUpdate` re-installs
+  **only on a formal release**: Claude Code refreshes when the version string changes, and the version
+  only moves in a tagged release commit, so ordinary `main` commits never pull teammates onto
+  un-released work. Not written on vendored (`cp -r`/pip) installs — there's no marketplace to enable
+  from. README documents it as the recommended team install; `scaffold.md` + the `settings.json.tmpl`
+  `_README` carry the rationale.
+- **`bin/selftest.sh` §21** asserts the enablement block is documented, is valid JSON, targets the
+  real marketplace repo, and is release-gated (`autoUpdate`).
+
+## [3.1.0] — 2026-07-06
+
+Hardening release from two real end-to-end runs (a `/setup` session and a full analysis ticket). Fixes
+the plugin/pip install path bugs those runs surfaced, de-hardcodes the shipped adapters for a clean
+first run on any stack, adds a SQL-portability guardrail, and flips the deliverable-CSV default to
+commit-by-default with a PII opt-out. Also adds an **Obsidian graph layer** for browsing the ticket
+archive as a knowledge graph.
+
+### Fixed
+- **Kit-root vs project-root resolution** (the class behind a `/ship` crash on plugin/pip installs).
+  `bin/enrich_ticket.py` resolved its sibling scripts off the project dir (`repo_root() / "bin"`),
+  which doesn't exist when the kit and project diverge — now resolved from the script's own kit dir.
+  The two SessionStart/PostToolUse index hooks imported the renderer from the project's `bin/` (silent
+  fail-open on a plugin install) — now from `${CLAUDE_PLUGIN_ROOT}` / the hook's own kit dir.
+  `bin/verify_stack.sh` resolved adapters off the stack.yaml location (the project) — now off the kit
+  (`${CLAUDE_PLUGIN_ROOT}`, else the script's dir), with a project fallback for repo-vendored adapters.
+- **Self-maintaining index went stale on ticket creation.** `/ticket` scaffolds via Bash, which the
+  `Write|Edit` PostToolUse hook never sees, so a new ticket didn't appear in `INDEX.md` until a manual
+  run. `/ticket` now renders the index as its final scaffold step; `build_ticket_index.py` also gained
+  a `git rev-parse --show-toplevel` fallback for by-hand runs without `CLAUDE_PROJECT_DIR`.
+
+### Changed
+- **Deliverable exports are committed by default.** The old `.gitignore` blanket-ignored
+  `**/final_deliverables/*.csv` (+ `.tsv`/`.xlsx`/`.parquet`), so results silently never reached git or
+  the PR. Now they're committed; PII/customer data opts out via a `*.private.csv` name or a `private/`
+  subfolder (both gitignored). `/ship` lists the deliverable files and asks you to confirm none carry
+  PII before committing. To restore the strict "docstore only" default, uncomment the four blanket
+  rules in the rendered `.gitignore` (documented in-file).
+- **Adapters de-hardcoded for a clean first run on any stack.** MCP server names now use the `{mcp}`
+  token from `stack.yaml` (was `mcp__atlassian__…` / `mcp__slack__…` / `mcp__plugin_productivity_*__…`);
+  the Jira and Azure DevOps `verify` no longer depend on the nullable `default_epic`; removed
+  org-specific Jira issue-type / mandatory-Epic / terminal-state content; `--parent` is now conditional
+  on `default_epic` being set. Added `mcp` to the asana/linear/monday adapter `requires`.
+- **`/setup` is install-mode aware.** On a plugin install it omits the settings.json `hooks` block
+  (`plugin.json` already wires them — duplicating double-fired), copies `statusline.sh` into the repo
+  so the statusline resolves, and offers to commit the scaffold at the end (so a later ticket PR
+  doesn't reference rules/adapters absent from history).
+- Ticket scaffolding no longer creates `.gitkeep` placeholders in subdirs that fill during build.
+- **Renamed the `commandify_everything` policy → `skillify_everything`**, completing the v2/v3
+  commands→skills shift: recurring work becomes a `/productize` **skill the agent can invoke itself**,
+  not a command a human must issue. Behavior is unchanged; update the key in an existing `stack.yaml`.
+- Spelled out **KISS (Keep It Simple, Stupid)** and **YAGNI (You Aren't Gonna Need It)** in the
+  always-loaded rules so a human reviewer knows exactly what they mean.
+
+### Added
+- **Obsidian graph layer.** `build_ticket_index.py` now also generates `tickets/graph/<id>.md`
+  (a node per ticket) and `tickets/objects/<object>.md` (a node per data object), so the repo opens as
+  an Obsidian vault: object clusters show every ticket that touched a table, and cross-refs show as
+  direct build-on lines. Plain markdown (no plugins/wikilinks), auto-maintained on the index hook, on
+  by default (`project.graph_notes: false` to disable). README convention untouched.
+- **Portable-params guardrail (CTE vs session `DECLARE`).** BigQuery/Snowflake/Synapse `dialect_notes`,
+  the `/review` dialect-lint tier, and `/spec-and-build` now steer to a `WITH params AS (…) … CROSS
+  JOIN params` row — a `DECLARE` script pollutes `--format csv` exports and breaks in single-statement
+  JDBC/ODBC clients (DataGrip/Simba).
+- **CSV deliverables use ASCII punctuation only.** Em/en dashes, smart quotes, and ellipsis characters
+  render as mojibake in Excel and many CSV viewers, so the always-loaded rules (`AGENTS.md`) and the
+  `/review` output-format check now steer cell values to plain ASCII (`-`, or "to" for ranges).
+- **Scaffolds a one-line `CLAUDE.md` (`@AGENTS.md`)** so Claude Code auto-loads the always-loaded
+  rules (it reads `CLAUDE.md`; other agents read `AGENTS.md`). `AGENTS.md` now also points at
+  `tickets/INDEX.md` / `OBJECTS.md`, putting the reuse-prior-work habit in the always-loaded tier.
+- **`bin/selftest.sh` §20** (path resolution + adapter hygiene) locks each fix above; §19 updated for
+  the new CSV default. Suite is at **119 checks**.
+
+## [3.0.0] — 2026-07-04
+
+Breaking: removed the 12 deprecated v1 alias stubs. The v2.0 release kept the old command names
+(`/start-ticket`, `/qc-review`, …) working as thin routers to their v2 skills, with a documented
+"removed in v3" lifecycle. That removal is this release. No engine, adapter, template, or skill
+behavior changed — purely the scheduled drop of the compatibility layer.
+
+### Removed
+- **12 v1 alias stubs** (`.claude/commands/`). Use the v2 names instead:
+  `start-ticket` / `prime-ticket` / `prime-domain` / `prime-warehouse` / `recall` → **`/ticket`**
+  (`--recall` for standalone lookups) · `qc-review` → **`/review`** · `deliver-ticket` →
+  **`/ship`** · `configure-workspace` / `onboard-teammate` → **`/setup`** (`--teammate`) ·
+  `productize-workflow` → **`/productize`** · `build-ticket-index` / `build-context-pack` →
+  **`/refresh`** (`index` / `context` / `all`). Full map: `docs/troubleshooting.md`.
+
+### Changed
+- `bin/selftest.sh` check 14b now asserts the 12 stubs are absent (was: asserts all present).
+- `.claude/hooks/session_context.py` dropped the now-unneeded deprecated-alias filter.
+
 ## [2.0.0] — 2026-07-01
 
 The UX release: **13 invokables → 7 skills**, one front door, ≤5-question setup, plain language on

{ticketwright-2.0.0 → ticketwright-3.2.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ticketwright
-Version: 2.0.0
+Version: 3.2.0
 Summary: Tool-agnostic AI layer for ticket-driven work repos: prior-art recall, object reverse-index, deep QC, and a self-maintaining ticket index. Stdlib-only, no vector store.
 Project-URL: Homepage, https://github.com/kyle-chalmers/ticketwright
 Project-URL: Repository, https://github.com/kyle-chalmers/ticketwright
@@ -59,6 +59,26 @@ That's it. `setup` also handles repos that **already have** ticket history — i
 existing layout instead of scaffolding, and writes a `MIGRATION.md` checklist (see
 [Adopting an existing repo](#adopting-an-existing-repo)).
 
+**Project-scoped by default.** A plugin can't set its own install scope — so instead, `setup` commits
+the enablement into the repo's `.claude/settings.json`. That's the *repo* opting in at project scope:
+it travels *with the repo*, so every teammate who opens (and trusts) the repo is prompted to install
+Ticketwright (no marketplace to add, no config to write), and it keeps working after the person who
+set it up moves on:
+
+```jsonc
+{
+  "extraKnownMarketplaces": {
+    "ticketwright": { "source": { "source": "github", "repo": "kyle-chalmers/ticketwright" }, "autoUpdate": true }
+  },
+  "enabledPlugins": { "ticketwright@ticketwright": true }
+}
+```
+
+`autoUpdate` re-installs **only on a formal release** — Claude Code refreshes when the plugin's
+*version* changes, and the version only moves in a tagged release commit, so day-to-day commits to
+`main` never pull teammates onto un-released work. (Prefer the user-level `/plugin install` above for
+personal, cross-repo use; use the committed block when you want the whole team on it.)
+
 ## How work flows
 
 Four steps — **plan → build → check → ship** — and one command to remember:
@@ -80,8 +100,8 @@ Three supporting skills you'll reach for occasionally:
 | `/productize` | Turn a recurring workflow (quarterly pull, monthly report) into its own parameterized, golden-tested skill |
 
 Plugin skills are namespaced (`/ticketwright:ticket`); inside a configured repo the short names
-work too. v1 command names (`/start-ticket`, `/qc-review`, …) still route to their v2 equivalents
-for this release.
+work too. (The v1 command names — `/start-ticket`, `/qc-review`, … — were retired in v3; see the
+rename map in [docs/troubleshooting.md](docs/troubleshooting.md#upgrading).)
 
 ## Never rebuild what's been built
 
@@ -92,6 +112,17 @@ vector store) and writes a *reuse brief*: what to copy, which gotchas carry over
 this time. `tickets/OBJECTS.md` answers the reverse question — "which tickets touched `VW_X`?"
 Details: [docs/ticket-index.md](docs/ticket-index.md).
 
+## See it as a graph (Obsidian)
+
+Ticketwright also writes a small, auto-maintained graph layer under `tickets/` — `graph/<id>.md`
+(a node per ticket) and `objects/<object>.md` (a node per data object) — so you can open the repo as
+an [Obsidian](https://obsidian.md) vault and *browse* your work: open a table like `ANALYTICS.VW_LOAN`
+and its local graph is every ticket that touched it; open a ticket and you see the objects it touched
+plus the tickets it built on. Point Obsidian at the repo (or `tickets/`), open Graph view, and for the
+cleanest picture set the filter to `-README -INDEX -OBJECTS -AGENTS -CLAUDE` and add a color Group on
+`path:objects`. It's plain markdown (no plugins, no wikilinks) and renders on GitHub too. On by
+default; set `project.graph_notes: false` in `stack.yaml` to turn it off.
+
 ## Safety rails (on by default)
 
 - **DB writes ask first** — a hook inspects every warehouse command and prompts before anything
@@ -99,8 +130,9 @@ Details: [docs/ticket-index.md](docs/ticket-index.md).
 - **External posts hard-halt** — `/ship` prints exactly what it's about to post (tracker comment,
   chat message, PR) and waits for your explicit go.
 - **Chat defaults to draft** — you click send.
-- **Exports can't leak into git** — the shipped `.gitignore` blocks `final_deliverables/*.csv` at
-  any depth; deliverables go to your docstore, not the repo.
+- **Deliverables commit with the ticket, PII opts out** — exports are committed by default so results
+  show in the PR; keep customer data out of git by naming it `*.private.csv` or dropping it in a
+  `private/` subfolder, and `/ship` lists what it's about to commit so nothing sensitive slips in.
 - **Every assumption is written down** — the ticket README template enumerates them by category.
 
 ## Adopting an existing repo
@@ -131,7 +163,7 @@ ticketwright init                       # vendor the kit into a repo (plugin-fre
 - [docs/ticket-index.md](docs/ticket-index.md) — the ticket catalog + recall engine in depth.
 - [CONTRIBUTING.md](CONTRIBUTING.md) · [ROADMAP.md](ROADMAP.md) · [CHANGELOG.md](CHANGELOG.md)
 
-CI runs the 95-check self-test on every push; PyPI publishing is OIDC Trusted Publishing (no stored
+CI runs the full self-test on every push; PyPI publishing is OIDC Trusted Publishing (no stored
 tokens) — see [docs/pypi-setup.md](docs/pypi-setup.md).
 
 ## License