@kurrent/kcap 0.11.30 → 0.11.32

package/kcap/.claude-plugin/plugin.json

@@ -1,5 +1,5 @@
 {
   "name": "kcap",
-  "version": "0.11.30",
+  "version": "0.11.32",
   "description": "Records and visualizes Claude Code sessions via kcap CLI hooks"
 }

package/kcap/skills/agent-flows/SKILL.md

@@ -163,7 +163,7 @@ The server enforces per-run budgets; watch for these in tool error responses:
 - **`400` starting `daemon_outdated:`** — the daemon's kcap is too old to host flow participants. Relay the server error's remediation verbatim — it names the outdated daemon; the fix: update kcap (`npm i -g @kurrent/kcap`), then `kcap daemon restart --name <its-name>` (works for detached and service-managed daemons alike — a raw `daemon stop` deliberately refuses a service-managed one; `kcap daemon status` lists names, and `--when-idle` defers the restart if the daemon is busy).
 - **`409` containing `participant_unreachable`** — that role's agent is in an ambiguous liveness state (its daemon disconnected or is restarting) or a possibly-live prior agent was spotted and the server is refusing to relaunch until its absence is proven, so it won't guess whether it's still alive rather than risk a duplicate launch. This is retryable — do not close the flow. Retry the send shortly; the server relaunches automatically once absence is proven, or ask the user to stop the participant (dashboard/API) and then re-send to force a fresh relaunch.
 - **Reserved-alias reviewer-model errors** (only when you passed `model`): `reviewer_model_protocol_required` — the server/daemon doesn't support a model override yet; drop `model` and retry, or tell the user to update. `reviewer_model_unavailable` — no resolver for that vendor recognizes the model; ask for a different one or drop the override. `model_vendor_mismatch` — the model belongs to a different vendor than the one selected; ask the user which vendor they meant. `reviewer_model_safe_settlement_required` — the server can't safely reapply this model on a heal/relaunch; close the flow and start a new one. `reviewer_model_unpriceable` — the model has no resolvable pricing to budget-check; ask for a differently-named/priced model.
-- **A round result of `unclear` whose text is exactly `participant_died` or `participant_stopped`** — that role's agent crashed or was stopped mid-round. The run stays **open**: address the same role again with `send_to_participant` and it relaunches automatically — the fresh agent has **no memory of prior rounds**, so restate any context it needs in your message; its earlier spend still counts against the run budget. No need to close and restart the flow. Other roles are unaffected and remain addressable in the meantime.
+- **A round result of `unclear` whose text is exactly `participant_died`, `participant_stopped`, or `participant_parked`** — that role's agent crashed, was stopped, or was parked for resume mid-round. In every case the run stays **open** and you address the same role again with `send_to_participant` to continue. The difference is what the next round gets: for `participant_died`/`participant_stopped` a **fresh** agent relaunches with **no memory of prior rounds**, so restate any context it needs; for `participant_parked` the same agent **resumes with its prior context intact** (a resumable park frees the slot between rounds). Earlier spend still counts against the run budget. No need to close and restart the flow; other roles are unaffected and remain addressable. (An older server reports a park as `participant_stopped` — still a resubmit trigger, so this works across the rollout.)
 
 ## Workflow
 

package/kcap/skills/review-flows/SKILL.md

@@ -148,7 +148,7 @@ After applying the role-surface safety gate, if `start_review_flow` / `submit_re
 - **`client_upgrade_required`, `flow_client_protocol_required`, or `flow_client_protocol_unsupported`** — update kcap; reserved review aliases fail closed on stale clients.
 - **`reserved_review_alias_shape`** — an admin changed a reserved alias to an invalid participant shape; restore exactly one participant named `reviewer`.
 - **`409` containing `participant_unreachable`** — the reviewer's liveness is ambiguous right now: either its daemon disconnected/is restarting, or the server positively spotted a possibly-live prior agent and is refusing to relaunch until its absence is proven (never launching a duplicate reviewer beside a possibly-live one). This is retryable — **do not close the flow.** Submit the round again shortly with `submit_review_round`; the server relaunches the reviewer automatically once absence is proven (via the daemon reconnecting/reporting, or the agent's death being confirmed). If it keeps failing across many retries, that points at the daemon itself needing attention, not at the flow needing to be restarted — ask the user to check daemon status, or stop the reviewer (dashboard/API) to force a fresh relaunch.
-- **A round result of `unclear` whose text is exactly `participant_died` or `participant_stopped`** — the reviewer crashed or was stopped mid-round. The run stays open: submit another round with `submit_review_round` and it relaunches automatically — the fresh reviewer has **no memory of prior rounds**, so restate any context it needs in your updated context; its earlier spend still counts against the run budget. No need to close and restart the flow.
+- **A round result of `unclear` whose text is exactly `participant_died`, `participant_stopped`, or `participant_parked`** — the reviewer crashed, was stopped, or was parked for resume mid-round. In every case the run stays open and you **submit another round** with `submit_review_round` to continue — no need to close and restart the flow. The difference is what the next round gets: for `participant_died`/`participant_stopped` a **fresh** reviewer relaunches with **no memory of prior rounds**, so restate any context it needs; for `participant_parked` the same reviewer **resumes with its prior context intact** (a resumable park frees the slot between rounds). Earlier spend still counts against the run budget. (An older server that predates the park signal reports a park as `participant_stopped` — still a resubmit trigger, so this works across the rollout.)
 
 ## Workflow
 

package/kcap/skills/suggest-review-flow/SKILL.md

@@ -0,0 +1,94 @@
+---
+name: suggest-review-flow
+description: >-
+  Use this skill when implementation is complete — a feature or bugfix is functionally
+  finished, tests pass, or you are about to commit / open a PR / merge — or when a spec is
+  finalized. At that moment, proactively OFFER an independent second-harness review even if
+  the user did not ask, because a fresh, separate reviewer catches what the author's own model
+  misses. It calls list_reviewer_vendors to recommend a reviewer that will actually run for
+  this repo, offers a spec-review or code-review flow, and hands off to the review-flows skill
+  only after the user accepts — it never starts a flow itself. Do NOT use it when the user
+  asks you to review something yourself ("review this", "code review this", "look over my
+  spec"), or mid-task before the work is finished.
+---
+
+# Suggest a review flow
+
+You have reached a milestone where an **independent** review adds real value: a
+separate coding-agent session, with none of this session's context, reviews the
+work — and a different model catches what the author's own model glossed over.
+Most people never think to do this. Your job here is to **offer** it at the
+right moment, recommend a reviewer that will actually run, and — only if the
+user says yes — hand off to the `review-flows` skill. **This skill never starts
+a review flow itself.**
+
+## When to offer (and when not to)
+
+Offer at exactly two milestones:
+
+- **Spec complete** — you just produced AND presented a finalized spec/design
+  this session → offer a `spec-review` flow.
+- **Implementation complete** — the change is functionally finished (you say so,
+  or you are about to commit / open a PR / merge) → offer a `code-review` flow.
+
+Do NOT offer (stay silent) when:
+
+- The user asked you to **review something yourself** ("review this", "code
+  review this", "look over the spec"). Just perform that review locally — no
+  offer, no flow.
+- You are only part-way through: rough mid-stream brainstorming, a partial plan,
+  a routine edit, or work whose known checks are still failing.
+- A review flow for this artifact is already engaged, active, or just completed.
+- Another skill (brainstorming, writing-plans, finishing-a-development-branch,
+  requesting-code-review) is mid-flow — let it finish; offer at the seam.
+
+**Offer at most once per artifact.** Key the offer on (milestone type + the
+target's identity + its revision): the same spec or the same change never gets
+offered twice, but a materially revised artifact may be offered again, and a
+second, different artifact of the same kind may be offered on its own. If the
+user declines or moves on, stay silent for that artifact. A failed availability
+lookup counts as the one prompt for that artifact — do not re-prompt.
+
+## How to recommend a reviewer
+
+1. Call **`list_reviewer_vendors`** (read-only; safe to call before offering).
+   It returns `reviewers[]` (vendors that can actually run for this repo now),
+   `driver_vendor` (the harness you are running in, or absent if unknown), and
+   `diagnostics`.
+2. **Prefer a reviewer different from the driver.** A different model is the
+   highest-value review. If one non-driver reviewer is available, recommend it;
+   if several, offer a short choice — never assert one vendor is "better".
+3. **If the only available reviewer is your own vendor, still offer it** —
+   framed as an independent, context-free reviewer session. The value is the
+   clean second look, not necessarily a different model.
+4. **If `driver_vendor` is absent (unknown)**, do NOT claim "a different model";
+   frame the offer as "a separate, independent reviewer session" and offer a
+   choice from `reviewers[]`.
+5. **If `reviewers[]` is empty**, read `diagnostics.reason` and give one line of
+   guidance instead of a dead end:
+   - `no_daemons_connected` → start a daemon (`kcap daemon start -d`).
+   - `no_repo_hosting_daemon` → run a daemon on a machine that has this repo.
+   - `no_unattended_reviewer` → install/certify a second harness as a reviewer.
+   - `repo_unresolved` → the current directory isn't a recognized repo.
+   - `lookup_failed` → couldn't reach the server; suggest checking `kcap status`.
+   - `schema_skew` → the kcap client/server is out of date; suggest updating.
+
+## Handing off (only after the user accepts)
+
+- You MAY run the read-only `list_reviewer_vendors` lookup automatically, but you
+  **MUST NOT start a review flow until the user affirmatively accepts THIS
+  offer.**
+- On acceptance, invoke the **`review-flows`** skill to run it, preserving the
+  chosen `kind` (spec-review / code-review), the target, and the chosen reviewer
+  vendor.
+- If the reviewer vendor turns out to be unavailable at start (a snapshot race),
+  `review-flows` surfaces the reason — do not silently substitute another vendor;
+  re-recommend and get fresh consent.
+
+## If `list_reviewer_vendors` is not available
+
+Your harness caches the kcap MCP tool list when it connects. If kcap was
+upgraded mid-session, `list_reviewer_vendors` may be missing from this session.
+Do NOT guess availability and do NOT shell out as a workaround — tell the user
+to restart the harness (or reconnect the `kcap-flows` MCP server) so the tool
+appears, then offer again.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@kurrent/kcap",
-  "version": "0.11.30",
+  "version": "0.11.32",
   "description": "CLI companion for Kurrent Capacitor — records and visualizes Claude Code sessions",
   "license": "SEE LICENSE IN LICENSE.md",
   "repository": {
@@ -14,17 +14,17 @@
     "postinstall": "node bin/postinstall.js"
   },
   "optionalDependencies": {
-    "@kurrent/kcap-darwin-arm64": "0.11.30",
-    "@kurrent/kcap-linux-arm64": "0.11.30",
-    "@kurrent/kcap-linux-musl-arm64": "0.11.30",
-    "@kurrent/kcap-linux-musl-x64": "0.11.30",
-    "@kurrent/kcap-linux-x64": "0.11.30",
-    "@kurrent/kcap-win-x64": "0.11.30"
+    "@kurrent/kcap-darwin-arm64": "0.11.32",
+    "@kurrent/kcap-linux-arm64": "0.11.32",
+    "@kurrent/kcap-linux-musl-arm64": "0.11.32",
+    "@kurrent/kcap-linux-musl-x64": "0.11.32",
+    "@kurrent/kcap-linux-x64": "0.11.32",
+    "@kurrent/kcap-win-x64": "0.11.32"
   },
   "files": [
     "bin/",
     "kcap/"
   ],
-  "kcapReleaseCommit": "73457022060d4d21d30052592425e1c1f020827c",
-  "kcapReleaseTag": "v0.11.30"
+  "kcapReleaseCommit": "e5b1eae185076df2ed3d6fc5914ad5648ea91ab2",
+  "kcapReleaseTag": "v0.11.32"
 }