@webpresso/codex-plugin 3.1.7 → 3.1.9

package/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "agent-kit",
-  "version": "3.1.7",
+  "version": "3.1.9",
   "description": "Webpresso agent-kit: blueprints, skills, hooks, and MCP server",
   "author": {
     "name": "Webpresso",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@webpresso/codex-plugin",
-  "version": "3.1.7",
+  "version": "3.1.9",
   "private": false,
   "description": "Codex plugin adapter for Webpresso agent-kit skills, hooks, and MCP runtime.",
   "homepage": "https://github.com/webpresso/agent-kit#readme",

package/skills/autopilot/SKILL.md

@@ -0,0 +1,39 @@
+---
+type: skill
+slug: autopilot
+title: Autopilot
+status: active
+scope: repo
+applies_to: [agents]
+related: [goal, plan-refine, verify, testing-philosophy]
+created: "2026-07-10"
+last_reviewed: "2026-07-10"
+name: autopilot
+description: "Drive a brief through the bounded native goal pipeline."
+argument-hint: "<brief|force: brief>"
+---
+
+# Autopilot
+
+Drive an approved brief through `plan -> execute -> qa -> validate -> land -> complete` using `wp goal run` and the current handoff.
+
+## Vague gate
+
+A brief is anchored when it includes a file path, symbol, issue number, test name, or numbered steps. Route any unanchored brief through `plan-refine` before execution. A leading `force:` is the only explicit bypass; strip it before creating the goal and record the bypass in the blueprint.
+
+## Pipeline
+
+1. Plan: refine blueprints and satisfy provenance-backed promotion approvals.
+2. Execute: implement current ready tasks; use only `wp_blueprint_task_verify` to mark them done.
+3. QA: run affected checks and `wp qa`; preserve normalized failure evidence.
+4. Validate: obtain distinct verification and code-review evidence and resolve findings. Each lane appends a tracked `reviews.md` record in one of these exact forms:
+
+   ```text
+   <!-- wp:goal-validation {"kind":"verify","reviewer":"<identity>","artifact":"<tracked-path>"} -->
+   <!-- wp:goal-validation {"kind":"code-review","reviewer":"<different-identity>","artifact":"<tracked-path>"} -->
+   ```
+
+5. Land: require green PR checks, complete the blueprint in the same PR, merge, and run owner cleanup.
+6. Complete: stop only when `wp goal run` records terminal evidence.
+
+Resume from durable state after interruption. Obey `WP_GOAL_DISABLE=1`, iteration/error/validation caps, and repository ownership. The pipeline must never self-approve, directly mark tasks done, bypass evidence, or push to main. Keep every loop bounded and never increase timeouts.

package/skills/goal/SKILL.md

@@ -0,0 +1,27 @@
+---
+type: skill
+slug: goal
+title: Goal
+status: active
+scope: repo
+applies_to: [agents]
+related: [autopilot, plan-refine, verify]
+created: "2026-07-10"
+last_reviewed: "2026-07-10"
+name: goal
+description: "Run durable multi-blueprint goals through evidence-gated handoffs."
+argument-hint: "<new|status|handoff|run|cancel> [brief]"
+---
+
+# Goal
+
+Use `wp goal` as the durable controller for work spanning one or more blueprints.
+
+## Protocol
+
+1. Create or resume the goal in a managed blueprint owner worktree.
+2. Read `wp goal status`, execute only the current phase handoff, then call `wp goal run` again.
+3. Complete blueprint tasks only through `wp_blueprint_task_verify` with passing evidence.
+4. Stop when the controller reports `complete`, `cancelled`, blocked guard exhaustion, or `WP_GOAL_DISABLE=1`.
+
+The blueprint engine owns task status and approval gates. The goal controller must never self-approve, directly mark a task done, bypass review provenance, or mutate the primary checkout. Keep every loop bounded and never increase timeouts.

package/skills/opencode-go/SKILL.md

@@ -44,46 +44,56 @@ MODEL=$(echo "$CATALOG" | grep '^opencode-go/kimi' | grep -- '-code$' | sort -V
 [ -z "$MODEL" ] && MODEL=$(echo "$CATALOG" | grep '^opencode-go/mimo' | grep -- '-pro$' | sort -V | tail -1)
 [ -z "$MODEL" ] && MODEL=$(echo "$CATALOG" | sort -V | tail -1)
 
-# Fail fast on blocked quota before launching a long review.
+# Fail fast on blocked quota before launching a long review. The probe uses
+# an idle timeout, not a wall-clock timeout, so active output is allowed to continue.
+# The probe succeeds as soon as it observes the expected success marker instead of
+# waiting for the CLI to finish teardown after the reply has already arrived.
 [ -n "$MODEL" ] || { echo "No OpenCode Go model resolved for this reviewer." >&2; exit 2; }
 OPENCODE_GO_REVIEW_EFFORT=${OPENCODE_GO_REVIEW_EFFORT:-medium}
 [ "$OPENCODE_GO_REVIEW_EFFORT" = medium ] || [ "$OPENCODE_GO_REVIEW_EFFORT" = high ] || { echo "OPENCODE_GO_REVIEW_EFFORT must be one of: medium, high" >&2; exit 2; }
 OPENCODE_GO_PROBE_TIMEOUT=${OPENCODE_GO_PROBE_TIMEOUT:-20}
+case "$OPENCODE_GO_PROBE_TIMEOUT" in
+  ""|*[!0-9]*) echo "OPENCODE_GO_PROBE_TIMEOUT must be a positive integer number of seconds" >&2; exit 2 ;;
+esac
+[ "$OPENCODE_GO_PROBE_TIMEOUT" -gt 0 ] || { echo "OPENCODE_GO_PROBE_TIMEOUT must be a positive integer number of seconds" >&2; exit 2; }
 PROBE_LOG=$(mktemp -t wp-opencode-go-probe.XXXXXX)
-PROBE_RC=0
-node - "$MODEL" "$OPENCODE_GO_PROBE_TIMEOUT" "$PROBE_LOG" "$OPENCODE_GO_REVIEW_EFFORT" <<'JS' || PROBE_RC=$?
-const { spawnSync } = require('node:child_process')
-const { writeFileSync } = require('node:fs')
-
-const [model, timeoutSecondsRaw, logPath, reviewEffort] = process.argv.slice(2)
-const timeoutSeconds = Number(timeoutSecondsRaw)
-if (!Number.isFinite(timeoutSeconds) || timeoutSeconds <= 0) {
-  console.error('OPENCODE_GO_PROBE_TIMEOUT must be a positive number of seconds')
-  process.exit(2)
-}
-
-const result = spawnSync(
-  'opencode',
-  ['run', '--format', 'json', '--variant', reviewEffort, '--model', model, '--title', 'opencode-go-credit-probe', 'Reply exactly OPENCODE_GO_OK.'],
-  { encoding: 'utf8', timeout: timeoutSeconds * 1000, maxBuffer: 1024 * 1024 },
-)
-
-let output = ''
-if (result.stdout) output += result.stdout
-if (result.stderr) output += result.stderr
-if (result.error?.code === 'ETIMEDOUT') {
-  output += `\nOPENCODE_GO_PROBE_TIMEOUT after ${timeoutSeconds}s\n`
-  writeFileSync(logPath, output)
-  process.exit(124)
-}
-if (result.error) {
-  output += `${result.error.message}\n`
-  writeFileSync(logPath, output)
-  process.exit(1)
-}
-writeFileSync(logPath, output)
-process.exit(typeof result.status === 'number' ? result.status : 1)
-JS
+unset PROBE_RC
+opencode run --format json --variant "$OPENCODE_GO_REVIEW_EFFORT" --model "$MODEL" --title opencode-go-credit-probe 'Reply exactly OPENCODE_GO_OK.' >"$PROBE_LOG" 2>&1 &
+PROBE_PID=$!
+LAST_PROBE_SIZE=$(wc -c <"$PROBE_LOG" | tr -d " ")
+LAST_PROBE_PROGRESS_AT=$(date +%s)
+while kill -0 "$PROBE_PID" 2>/dev/null; do
+  if grep -q "OPENCODE_GO_OK" "$PROBE_LOG"; then
+    kill "$PROBE_PID" 2>/dev/null || true
+    wait "$PROBE_PID" 2>/dev/null || true
+    PROBE_RC=0
+    break
+  fi
+  CURRENT_PROBE_SIZE=$(wc -c <"$PROBE_LOG" | tr -d " ")
+  NOW=$(date +%s)
+  if [ "$CURRENT_PROBE_SIZE" -gt "$LAST_PROBE_SIZE" ]; then
+    LAST_PROBE_SIZE=$CURRENT_PROBE_SIZE
+    LAST_PROBE_PROGRESS_AT=$NOW
+  elif [ $((NOW - LAST_PROBE_PROGRESS_AT)) -gt "$OPENCODE_GO_PROBE_TIMEOUT" ]; then
+    printf '\nOPENCODE_GO_PROBE_TIMEOUT after %ss without progress\n' "$OPENCODE_GO_PROBE_TIMEOUT" >>"$PROBE_LOG"
+    kill -KILL "$PROBE_PID" 2>/dev/null || true
+    wait "$PROBE_PID" 2>/dev/null || true
+    PROBE_RC=124
+    break
+  fi
+  sleep 1
+done
+if [ -z "${PROBE_RC+x}" ]; then
+  if grep -q "OPENCODE_GO_OK" "$PROBE_LOG"; then
+    PROBE_RC=0
+  else
+    wait "$PROBE_PID" || PROBE_RC=$?
+  fi
+fi
+if [ "$PROBE_RC" -eq 0 ] && ! grep -q "OPENCODE_GO_OK" "$PROBE_LOG"; then
+  echo "OpenCode Go probe finished without the expected success marker." >>"$PROBE_LOG"
+  PROBE_RC=1
+fi
 if [ "$PROBE_RC" -ne 0 ]; then
   echo "OpenCode Go probe failed or timed out for $MODEL after ${OPENCODE_GO_PROBE_TIMEOUT}s; skipping this reviewer instead of starting the full review." >&2
   sed -n '1,80p' "$PROBE_LOG" >&2

package/skills/verify/SKILL.md

@@ -162,7 +162,7 @@ user explicitly asked for a different approval count or reviewer mix.
 
 Default policy:
 
-- Require **two extra model approvals** by default.
+- Require **two extra model approvals** by default, but at most **one OpenCode Go reviewer** is required unless the user explicitly asks for multiple OpenCode models.
 - If the user asks for more, fewer, zero, or specific reviewers, follow that
   explicit instruction and report the chosen requirement.
 - These approvals are model outside-voice approvals, not a substitute for
@@ -172,10 +172,10 @@ Reviewer preference:
 
 - When running from Codex, prefer **Claude + one OpenCode Go reviewer**.
 - When running from Claude, prefer **Codex + one OpenCode Go reviewer**.
-- When Codex/Claude outside voice is unavailable, use **two distinct OpenCode Go
-  reviewers** if OpenCode is logged in.
+- When Codex/Claude outside voice is unavailable, use **one OpenCode Go
+  reviewer** if OpenCode is logged in.
 - Prefer the repo's outside-voice skills instead of hand-rolled commands:
-  `$agent-kit:claude`, `$agent-kit:codex`, and one or more OpenCode Go reviewer
+  `$agent-kit:claude`, `$agent-kit:codex`, and one OpenCode Go reviewer
   skills such as `$agent-kit:opencode-go`, `$agent-kit:qwen`,
   `$agent-kit:deepseek`, `$agent-kit:glm`, `$agent-kit:kimi`,
   `$agent-kit:minimax`, or `$agent-kit:mimo`.