@event4u/agent-config 2.18.0 → 2.20.0

package/.agent-src/commands/agent-status.md

@@ -41,6 +41,22 @@ Calculate based on these estimates:
 | **Matched skills** | count skills triggered this request × ~1,500 avg | variable |
 | **Platform overhead** | system prompt, tool schemas | ~15,000 |
 
+### 3a. Read session cost ledger (measured, not estimated)
+
+Run `node scripts/cost/track.mjs` (silent — `TRACK_QUIET=1`) and parse
+the last record from `agents/cost-tracking/sessions.jsonl`. If the file
+does not exist (tracker never run for this project), skip this step and
+note `cost ledger: not initialised` in the dashboard.
+
+Extract from latest record:
+
+- `total_usd` — dollars spent in current session
+- `by_model[]` — per-tier (haiku / sonnet / opus) input / output / cache split
+- `budget.tier` — `under` / `50` / `75` / `90` / `100` (from `node scripts/cost/budget.mjs check`)
+
+Pricing source: [`bench/pricing.yaml`](../../bench/pricing.yaml). Reader
+implementation: [`scripts/cost/track.mjs`](../../scripts/cost/track.mjs).
+
 ### 4. Calculate freshness thresholds
 
 - **Message threshold**: Next multiple of 25 ≥ current count
@@ -70,6 +86,19 @@ Use Markdown tables and headings — NOT ASCII box art (breaks in non-monospace
 | Matched skills (~{n}) | ~{n×1500} |
 | **Total input** | **~{sum}** |
 
+**💵 Session cost (measured)**
+
+If ledger exists, render:
+
+| | |
+|---|---|
+| 💵 Session total | ${total_usd} |
+| Haiku / Sonnet / Opus | ${haiku} / ${sonnet} / ${opus} |
+| 🎯 Budget tier | {emoji} {tier} ({utilization_pct}% of cap) |
+
+Tier-emoji map: `under` / `50` → ✅ · `75` → ⚠️ · `90` → ⚠️⚠️ · `100` → ❌.
+If ledger does not exist, render `Session cost: not initialised — run \`task cost:track\` to start measuring`.
+
 **⚡ Freshness**
 
 | | |

package/.agent-src/commands/onboard.md

@@ -13,48 +13,51 @@ suggestion:
 
 # /onboard
 
-Centralized first-run flow. Bundles scattered "ask once" prompts (user_name,
-IDE, rtk install, cost profile, learning loop) into one interactive setup.
-Ends by setting `onboarding.onboarded: true` in `.agent-settings.yml`.
+Centralized first-run flow. Bundles what used to be scattered "ask once"
+prompts (user_name, IDE, rtk install, cost profile, learning loop) into a
+single interactive setup. Ends by setting `onboarding.onboarded: true` in
+`.agent-settings.yml`.
 
-Triggered by [`onboarding-gate`](../rules/onboarding-gate.md) when
-`onboarding.onboarded` is `false`, or by explicit re-run.
+Triggered by the [`onboarding-gate`](../rules/onboarding-gate.md) rule when
+`onboarding.onboarded` is `false` or by the user explicitly re-running it.
 
 ## When NOT to use
 
 - Change cost profile only → [`/set-cost-profile`](set-cost-profile.md).
-- Single-value edit → ask agent to change it, or edit `.agent-settings.yml`
-  directly per [`layered-settings`](../docs/guidelines/agent-infra/layered-settings.md).
+- Single-value edit → ask the agent to change it, or edit
+  `.agent-settings.yml` directly. The agent follows the merge rules in
+  [`layered-settings`](../docs/guidelines/agent-infra/layered-settings.md).
 
 ## Preconditions
 
-`.agent-settings.yml` exists. If missing, tell user to run `scripts/install`
-(or `python3 scripts/install.py`) first and stop — command assumes file +
-template defaults are in place.
+`.agent-settings.yml` exists. If missing, tell the user to run
+`scripts/install` (or `python3 scripts/install.py`) first and stop — this
+command assumes the file and its template-derived defaults are in place.
 
 ## Steps
 
 ### 1. Greet and set expectations
 
-One line: one-time setup, six questions, one at a time (iron law from
-`user-interaction`).
+Keep it short. One line explaining this is the one-time setup, up to
+nine questions, one at a time, following the iron law (`user-interaction`).
 
 ### 2. Offer user-global cross-project defaults
 
-Detect whether `~/.event4u/agent-config/agent-settings.yml` exists (or
-legacy `~/.config/agent-config/agent-settings.yml`, read as fallback by
-every loader). New path namespaces every package-owned user-global
-artefact under one root — same place `anthropic.key`, `openai.key`,
-`council-spend.jsonl` now live.
-
-- **File exists** → skip step entirely. Re-onboarding never overwrites
-  user-global file silently.
-- **File missing AND first-time setup heuristic** — heuristic for "first
-  machine setup": no other `.agent-settings.yml` in any sibling project on
-  disk. Conservative shell probe:
+Detect whether `~/.event4u/agent-config/agent-settings.yml` already
+exists (or the legacy `~/.config/agent-config/agent-settings.yml`,
+which is read as a fallback by every loader). The new path namespaces
+every package-owned user-global artefact under one root — same place
+where `anthropic.key`, `openai.key`, and `council-spend.jsonl` now
+live.
+
+- **File exists** → skip this step entirely. Re-onboarding never
+  overwrites a user-global file silently.
+- **File missing AND first-time setup heuristic** — heuristic for
+  "first machine setup": no other `.agent-settings.yml` in any
+  sibling project on disk. Conservative shell probe:
   `find $(dirname "$PWD") -maxdepth 3 -name .agent-settings.yml 2>/dev/null | grep -v "^$PWD/" | head -1`
-  → non-empty → developer done this before, **skip**.
-  → empty → first-time setup, ask:
+  → non-empty means the developer has done this before, **skip**.
+  → empty means first-time setup, ask:
 
 ```
 > A user-global config at ~/.event4u/agent-config/agent-settings.yml lets
@@ -69,12 +72,13 @@ artefact under one root — same place `anthropic.key`, `openai.key`,
 > 2. No — keep settings project-local only
 ```
 
-If user picks `1`, **defer write** to tail step (see step 9). Capture choice
-in working memory only; do **not** create file here. File gets written
-**after** project-local values are confirmed, so initial values mirror
-what developer just chose for this project.
+If the user picks `1`, **defer the write** to a tail step
+(see step 9 below). Capture the choice in working memory only; do
+**not** create the file here. The file gets written **after** the
+project-local values are confirmed, so its initial values mirror
+what the developer just chose for this project.
 
-If user picks `2`, set working-memory flag to skip step 9.
+If the user picks `2`, set a working-memory flag to skip step 9.
 
 ### 3. Capture `personal.user_name`
 
@@ -87,11 +91,11 @@ Skip if already set (non-empty). Otherwise:
 > 2. Skip — stay anonymous
 ```
 
-Free-text → write to `personal.user_name`. `2` → leave empty.
+Free-text answer → write to `personal.user_name`. `2` → leave empty.
 
 ### 4. Capture `personal.ide` (with auto-detect)
 
-Skip if set. Otherwise auto-detect first:
+Skip if already set. Otherwise auto-detect first:
 
 ```bash
 ps aux | grep -iE '(Visual Studio Code|Code Helper|phpstorm|cursor)' | grep -v grep
@@ -109,13 +113,13 @@ ps aux | grep -iE '(Visual Studio Code|Code Helper|phpstorm|cursor)' | grep -v g
 > 4. Skip — I'll configure it later
 ```
 
-If IDE set, also ask `personal.open_edited_files` (`true`/`false`).
+If IDE is set, also ask about `personal.open_edited_files` (`true`/`false`).
 
 ### 5. Capture `personal.pr_comment_bot_icon`
 
-Personal preference — each developer decides how own PR replies look. Skip
-only if user already set non-default deliberately (agent can't tell, so
-always ask on first run):
+Personal preference — each developer decides how their own PR replies
+should look. Skip only if the user has already set a non-default value
+deliberately (agent can't tell, so always ask on first run):
 
 ```
 > When I reply to PR review comments on your behalf, should I prefix each
@@ -144,17 +148,16 @@ Silent `which rtk`.
 ```
 
 `1` or `2` → run install, on success set `rtk_installed: true` and apply
-rtk post-install steps (telemetry off, init --global) per
-[`rtk-output-filtering`](../skills/rtk-output-filtering/SKILL.md).
-`3` → leave `rtk_installed: false`, move on. No "ask again tomorrow" —
-`/onboard` is one-shot.
-
+rtk post-install steps (telemetry off, init --global) per the
+[`rtk-output-filtering`](../skills/rtk-output-filtering/SKILL.md) skill.
+`3` → leave `rtk_installed: false` and move on. No "ask again tomorrow"
+logic — `/onboard` is one-shot.
 
 ### 7. Confirm `cost_profile` and learning loop
 
 Read current `cost_profile` and `pipelines.skill_improvement` values.
-Present plainly (sensible defaults from template — `balanced` +
-`skill_improvement: true`; rationale in
+Present them plainly (they already have sensible defaults from the
+template — `balanced` + `skill_improvement: true`; rationale lives in
 [`docs/contracts/cost-profile-defaults.md`](../docs/contracts/cost-profile-defaults.md)):
 
 ```
@@ -166,19 +169,148 @@ Present plainly (sensible defaults from template — `balanced` +
 > 3. Disable learning loop — sets pipelines.skill_improvement=false
 ```
 
-`2` → defer to `/set-cost-profile` and return here. `3` → flip toggle.
+`2` → defer to `/set-cost-profile` and return here. `3` → flip the toggle.
+
+### 7a. Capture `user_type` and `profile.id` (role selection)
+
+Skip if `profile.id` is already set in `.agent-settings.yml` — re-running
+`/onboard` never silently overwrites a previously chosen role. If unset,
+ask:
+
+```
+> Which audience fits you best? This picks the audience profile that
+> shapes which skills, rules, and MCP servers the agent surfaces.
+>
+> 1. Software — coder, engineer, technical lead         → developer
+> 2. Content — writer, creator, editorial               → content_creator
+> 3. Founder — solo founder, early-stage CEO            → founder
+> 4. Consulting — agency, freelance, client work        → agency
+> 5. Marketing — campaigns, growth, brand               → content_creator
+> 6. Finance — bookkeeping, controlling, CFO            → finance
+> 7. Handwerk — operations, trades, admin               → ops
+> 8. Self-configure — leave profile.id empty, I'll edit .agent-settings.yml manually
+```
+
+Map answer → write **both** keys to `.agent-settings.yml` (one block,
+section-aware merge):
+
+- `personal.user_type`: the label as `"software" | "content" | "founder"
+  | "consulting" | "marketing" | "finance" | "handwerk" | "self_configure"`
+  (stable audit field; never auto-mutated).
+- `profile.id`: the profile-loader id from the right column. Choice `8`
+  → leave `profile.id` unset (loader falls back to its default chain
+  documented in [`docs/contracts/profile-system.md`](../docs/contracts/profile-system.md)).
+
+Six profile YAMLs ship today (`developer` · `content_creator` · `founder`
+· `agency` · `finance` · `ops`) — Marketing collapses into
+`content_creator` per the closest-audience mapping. Verify the chosen
+id exists at
+[`.agent-src.uncompressed/profiles/<id>.yml`](../../.agent-src.uncompressed/profiles/)
+before writing; missing file → fall back to `self_configure` and surface
+the path so the user can author it later.
+
+### 7b. Stack confirmation
+
+Skip if `stack.detected` is already non-empty (re-runnable). Otherwise
+run the offline shell probe and present the result for confirmation —
+no network call, file-existence only:
+
+```bash
+stacks=()
+[ -f composer.json ] && stacks+=("php")
+[ -f package.json ] && stacks+=("node")
+[ -f Cargo.toml ] && stacks+=("rust")
+[ -f go.mod ] && stacks+=("go")
+[ -f pyproject.toml ] || [ -f requirements.txt ] && stacks+=("python")
+[ -f Gemfile ] && stacks+=("ruby")
+```
+
+Then:
+
+- **One stack found** → confirm:
+  `> Detected {stack}. 1. Yes, use it  2. Override (pick from list)  3. Skip — I'll set stack.detected later`
+- **Multiple stacks found** → list them:
+  `> Detected {a, b}. 1. Use all  2. Pick primary  3. Skip`
+- **None found** → ask:
+  ```
+  > No stack manifest found in this project root. Pick the primary stack
+  > so profile-aware presets can resolve (this is advisory; the agent
+  > re-detects on each session):
+  >
+  > 1. php          2. node         3. rust
+  > 4. go           5. python       6. ruby
+  > 7. Skip — leave stack.detected empty
+  ```
+
+Write the result to `.agent-settings.yml`:
+
+```yaml
+stack:
+  detected: ["php"]              # or ["php","node"], or [] for skip
+  source: "wizard"               # audit field: wizard | probe | manual
+```
+
+### 7c. Risk-appetite (`preset.id`)
+
+Skip if `preset.id` is already set. Otherwise ask:
+
+```
+> Risk appetite for autonomy, cost caps, and council usage:
+>
+> 1. fast      — minimal floors, higher daily caps, council auto-consult on
+> 2. balanced  — default — moderate caps, council on-demand
+> 3. strict    — tight caps, council disabled, high-confidence required
+>
+> Recommendation: 2 (balanced).
+>
+> See docs/contracts/config-presets.md for the per-knob comparison.
+```
+
+Map answer → write `preset.id: "fast" | "balanced" | "strict"` to
+`.agent-settings.yml`. Default fallback (no answer / cloud surface) is
+`balanced`, mirroring the loader default at
+[`scripts/config/presets.py`](../../scripts/config/presets.py).
 
 ### 8. Mark onboarded
 
-Write `onboarding.onboarded: true` to `.agent-settings.yml` using
+Write `onboarding.onboarded: true` to `.agent-settings.yml` using the
 section-aware merge rules from
 [`layered-settings`](../docs/guidelines/agent-infra/layered-settings.md#section-aware-merge-rules)
-(preserve comments, key order, touch only changed fields).
+(preserve comments, key order, touch only the changed fields).
+
+At this point the file holds the full wizard output:
+
+```yaml
+profile:
+  id: "developer"          # step 7a (omitted if user picked Self-configure)
+preset:
+  id: "balanced"           # step 7c
+personal:
+  user_type: "software"    # step 7a — stable audit field
+  user_name: "..."
+  ide: "..."
+  # ...other personal.* from earlier steps
+stack:
+  detected: ["php"]        # step 7b
+  source: "wizard"
+onboarding:
+  onboarded: true          # this step
+```
+
+Verify the chain resolves before flipping `onboarded: true`:
+
+```bash
+./agent-config explain config --json | python3 -c "import json,sys; d=json.load(sys.stdin); print(d['profile']['id'], d['preset']['id'])"
+```
+
+Non-zero exit → surface the error, keep `onboarded: false`, ask the user
+to re-run `/onboard`. Zero exit → write `onboarded: true`.
 
 ### 9. Write user-global file (only if opted in at step 2)
 
-Skip unless step 2 captured explicit "yes". Re-confirm intent in one line —
-never silent-write a file outside project tree:
+Skip this step unless step 2 captured an explicit "yes" from the user.
+Re-confirm intent in one line — never silent-write a file outside the
+project tree:
 
 ```
 > Writing ~/.event4u/agent-config/agent-settings.yml with the six
@@ -196,16 +328,16 @@ never silent-write a file outside project tree:
 ```
 
 `1` → ensure `~/.event4u/agent-config/` exists (`mkdir -p`, mode `0700`;
-migration shim in `scripts/install.py` moves any legacy
-`~/.config/agent-config/` files into new namespace on first run),
-then write file with mode `0600`. Schema is **flat-or-nested YAML keyed on
-dotted paths** in whitelist documented in
+the migration shim in `scripts/install.py` moves any legacy
+`~/.config/agent-config/` files into the new namespace on first run),
+then write the file with mode `0600`. The schema is **flat-or-nested
+YAML keyed on the dotted paths** in the whitelist documented in
 [`scripts/_lib/agent_settings.py`](../../scripts/_lib/agent_settings.py).
-Use same section-aware merge rules from
+Use the same section-aware merge rules from
 [`layered-settings`](../docs/guidelines/agent-infra/layered-settings.md#section-aware-merge-rules)
-**only if file unexpectedly already exists** between step 2 and this step
-(race condition); otherwise create from scratch with exact six keys above
-and a one-line file header comment:
+**only if the file unexpectedly already exists** between step 2 and
+this step (race condition); otherwise create from scratch with the
+exact six keys above and a one-line file header comment:
 
 ```yaml
 # event4u/agent-config — user-global DX-comfort defaults
@@ -222,6 +354,10 @@ Echo what was captured, in one block:
 ```
 ✅  Onboarding complete.
 
+  profile.id: {value or — (Self-configure)}
+  preset.id: {value or balanced}
+  personal.user_type: {value}
+  stack.detected: {comma-list or —}
   personal.user_name: {value or —}
   personal.ide: {value or —}
   personal.open_edited_files: {value}
@@ -239,8 +375,8 @@ you ask it to change a value.
 
 ### 11. Quickstart pointer — next command after onboarding
 
-Print next-action block so developer does **not** have to re-find README
-to start work. One screen, no question, no prompt:
+Print the next-action block so the developer does **not** have to re-find
+the README to start work. One screen, no question, no prompt:
 
 ```
 🚀  Next step — your first real task
@@ -253,23 +389,23 @@ to start work. One screen, no question, no prompt:
         confirming the work-engine ran end-to-end.
 
     /implement-ticket PROJ-123
-        Ticket-driven flow — pulls Jira/Linear context, runs same
+        Ticket-driven flow — pulls Jira/Linear context, runs the same
         refine → memory → analyze → plan → implement → test → verify
         → report sequence, halts on ambiguity.
 
-  Both honour decision_engine gates in .agent-settings.yml
-  (see docs/contracts/decision-engine-gates.md for schema).
+  Both honour the decision_engine gates in .agent-settings.yml
+  (see docs/contracts/decision-engine-gates.md for the schema).
 ```
 
-Skip block in cloud surfaces (cloud agent's invocation surface is
-already chat window).
+Skip this block in cloud surfaces (the cloud agent's invocation surface
+is already the chat window).
 
 ### 12. Maintainer-only feature pointer
 
-Print one-screen hint after summary — no question, no prompt, just pointer
-for maintainers who want to opt into artefact-engagement telemetry layer.
-Consumers can ignore; feature is **default-off** and stays off unless
-explicitly enabled.
+Print a one-screen hint after the summary — no question, no prompt, just a
+pointer for maintainers who want to opt into the artefact-engagement
+telemetry layer. Consumers can ignore it; the feature is **default-off**
+and stays off unless explicitly enabled.
 
 ```
 ℹ️  Maintainer telemetry (opt-in)
@@ -287,28 +423,32 @@ Skip this block in cloud surfaces (no settings file, no log path).
 
 ## Gotchas
 
-- `.agent-settings.yml` is git-ignored. Command never commits.
-- One question per turn. Iron law from `ask-when-uncertain` applies; do
-  not stack questions 2–9 into single prompt.
+- `.agent-settings.yml` is git-ignored. This command never commits.
+- One question per turn. The iron law from `ask-when-uncertain` applies;
+  do not stack questions 2–9 into a single prompt.
 - Re-running `/onboard` when `onboarded: true` is allowed — walk through
-  all steps again and rewrite values user confirms.
-- Never overwrite non-empty value without asking (applies to `user_name`,
-  `ide`).
+  all steps again and rewrite the values the user confirms.
+- Never overwrite a non-empty value without asking (applies to `user_name`,
+  `ide`, `profile.id`, `preset.id`, `personal.user_type`, and
+  `stack.detected`).
+- **Offline-only.** Steps 7a / 7b / 7c never call out to the network.
+  Stack probe = file existence; profile / preset resolution = local YAML
+  reads only.
 - **User-global file is opt-in, one-shot, never silent.** Step 2 captures
-  intent, step 9 re-confirms before actual write. If
-  `~/.event4u/agent-config/agent-settings.yml` (or legacy
+  intent, step 9 re-confirms before the actual write. If
+  `~/.event4u/agent-config/agent-settings.yml` (or the legacy
   `~/.config/agent-config/agent-settings.yml`) already exists when
   `/onboard` starts, step 2 is skipped entirely — re-onboarding never
-  silently rewrites developer's cross-project defaults. Use
-  `/sync-agent-settings` (project-scoped only) or edit file manually for
-  mid-life changes.
+  silently rewrites a developer's cross-project defaults. Use
+  `/sync-agent-settings` (project-scoped only) or edit the file
+  manually for mid-life changes.
 
 ## Cloud Behavior
 
 On cloud surfaces (Claude.ai Web, Skills API) this command is **fully inert** —
-no `.agent-settings.yml` to write, no `onboarding.onboarded` key to flip,
-no local IDE/rtk env to capture. First-run setup is local-agent concern;
-cloud agent should proceed without invoking it.
+there is no `.agent-settings.yml` to write, no `onboarding.onboarded` key to
+flip, and no local IDE/rtk environment to capture. First-run setup is a
+local-agent concern; the cloud agent should proceed without invoking it.
 
 ## See also
 

package/.agent-src/commands/refine-ticket.md

@@ -77,6 +77,9 @@ persona voices) plus the close-prompt. Stop there. Do **not** chain into
 /refine-ticket PROJ-123
 /refine-ticket                              # uses current branch
 /refine-ticket --personas=+qa PROJ-123
+/refine-ticket --user-type=galabau-field-crew PROJ-123
+/refine-ticket --personas=+qa --user-type=truck-driver PROJ-123
+/refine-ticket --user-type=metalworking-shop PROJ-123
 /refine-ticket --fresh-eyes https://acme.atlassian.net/browse/PROJ-123
 ```
 

package/.agent-src/packs/README.md

@@ -0,0 +1,49 @@
+# Workflow packs
+
+Seed packs that bundle `(profile + preset + command-set +
+skill-allowlist + personas)` into a single opinionated stance. A
+user adopts a pack instead of picking five independent settings.
+
+**Schema:** [`docs/contracts/workflow-packs.md`](../../docs/contracts/workflow-packs.md).
+
+## Shipped packs
+
+| Pack id | Profile | Preset | One-liner |
+|---|---|---|---|
+| [`founder-mvp`](founder-mvp.yml) | `founder` | `fast` | Ship the MVP and the pitch deck in the same week. |
+| [`content-engine`](content-engine.yml) | `content_creator` | `balanced` | Editorial calendar, brand voice, and ghostwriter on one loop. |
+| [`agency-delivery`](agency-delivery.yml) | `agency` | `strict` | Multi-client refine → estimate → deliver with audit-grade trace. |
+
+## Activation
+
+```bash
+# Adopt a pack at install time:
+npx @event4u/agent-config init --pack founder-mvp
+
+# Adopt a pack later (writes pack.id to .agent-settings.yml):
+agent-config onboard --pack content-engine
+
+# Revert to the underlying profile + preset defaults:
+agent-config onboard --pack none
+```
+
+Packs are an **opt-in layer above the profile + preset chain**.
+Removing a pack reverts cleanly; no data is lost.
+
+## Constraints (enforced by the schema, not runtime)
+
+- `commands_allowed` ≤ 12, `skills_allowed` ≤ 15, `personas` ≤ 4.
+- Every referenced command / skill / persona / profile / preset
+  MUST resolve to an existing artefact.
+- Packs cannot widen safety floors. Domain-safety rules apply
+  unconditionally.
+
+## Adding a pack
+
+1. Pick a `<pack-id>` that does not collide with any profile or preset id.
+2. Copy one of the shipped packs as a template.
+3. Fill in `rationale.*` — every pack must justify its composition in
+   plain prose.
+4. Add the row to the table above.
+5. Open a PR; reviewer validates against the schema by hand until
+   `scripts/lint_packs.py` lands.

package/.agent-src/packs/agency-delivery.yml

@@ -0,0 +1,63 @@
+# Seed pack — agency-delivery
+# Schema: docs/contracts/workflow-packs.md
+pack:
+  id: agency-delivery
+  audience:
+    label: "Multi-client delivery shop with audit-grade decision trace"
+    one_liner: "Multi-client refine → estimate → deliver with audit-grade trace."
+  composition:
+    profile_id: agency
+    preset_id: strict
+  surface:
+    commands_allowed:
+      - work
+      - implement-ticket
+      - refine-ticket
+      - estimate-ticket
+      - feature
+      - roadmap
+      - review-changes
+      - create-pr
+      - commit
+      - council
+    skills_allowed:
+      - doc-coauthoring
+      - decision-record
+      - refine-ticket
+      - estimate-ticket
+      - perf-feedback-craft
+      - adr-create
+      - rice-prioritization
+      - launch-readiness
+      - verify-completion-evidence
+      - minimal-safe-diff
+      - test-driven-development
+      - api-design
+      - threat-modeling
+      - logging-monitoring
+      - dashboard-design
+    personas:
+      - reviewer
+      - product-strategist
+      - editor
+      - security
+  rationale:
+    why_this_profile: |
+      Agency profile pre-loads the documentation, decision-record,
+      and refine/estimate skills — the artefacts a multi-client shop
+      hands to a client at every phase boundary. The decision-record
+      anchor is load-bearing: without it, a year of client work
+      becomes opaque the moment the lead engineer rotates off.
+    why_this_preset: |
+      Strict preset is mandatory for agencies. Client work is
+      contractually scoped, so silently widening scope is a billing
+      risk; confidence floor and evidence-required mean every diff
+      can be defended at a client review. Council auto-consult is
+      on for the dollar-value bands that hit the contract.
+    why_these_commands: |
+      /refine-ticket and /estimate-ticket sit at the start of every
+      engagement; /implement-ticket and /review-changes drive
+      delivery; /create-pr standardises the handoff back to the
+      client's repo; /roadmap and /feature handle multi-phase
+      engagements; /council provides a second opinion at decision
+      gates billed to the client.

package/.agent-src/packs/content-engine.yml

@@ -0,0 +1,53 @@
+# Seed pack — content-engine
+# Schema: docs/contracts/workflow-packs.md
+pack:
+  id: content-engine
+  audience:
+    label: "Writers, ghostwriters, marketers running an editorial pipeline"
+    one_liner: "Editorial calendar, brand voice, and ghostwriter on one loop."
+  composition:
+    profile_id: content_creator
+    preset_id: balanced
+  surface:
+    commands_allowed:
+      - work
+      - post-as
+      - ghostwriter
+      - optimize-prompt
+      - refine-prompt
+      - challenge-me
+      - commit
+    skills_allowed:
+      - voice-and-tone-design
+      - messaging-architecture
+      - editorial-calendar
+      - release-comms
+      - prompt-engineering-patterns
+      - content-funnel-design
+      - discovery-interview
+      - voc-extract
+      - deep-reading-analyst
+      - doc-coauthoring
+    personas:
+      - editor
+      - brand-voice
+      - product-strategist
+  rationale:
+    why_this_profile: |
+      Content creator profile pre-loads voice, tone, and editorial
+      planning skills — the three things a content operator returns
+      to every day. Ghostwriting and post-as primitives are surfaced
+      at the top of the command list because they are the package's
+      distinctive non-developer affordance.
+    why_this_preset: |
+      Balanced preset matches the content workflow: most drafts are
+      reversible, most calls do not touch prod data, the cost of an
+      extra clarifying question is small relative to a misframed
+      piece. Council auto-consult stays off — content trade-offs
+      are usually taste, not models.
+    why_these_commands: |
+      /work and /post-as are the daily entry points. /ghostwriter
+      handles voice fetches and disclosure-footed drafts.
+      /optimize-prompt and /refine-prompt sharpen briefs before they
+      reach the writing skills. /challenge-me grills positioning
+      before launch. /commit ships finished drafts as repo artefacts.