[object Object]

← back to Cli Printing Press

fix(skill): force agent readiness review to run in foreground

feef5fa560a89d73f984cdcf78eda8fb4aeeefab · 2026-03-28 00:00:01 -0700 · Trevin Chow

The cli-agent-readiness-reviewer dispatch in Phase 4.9 was being
backgrounded, causing the skill to skip straight to Phase 5 without
waiting for results. Add explicit foreground-only instructions and
run_in_background: false to both the initial dispatch and the re-run.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Files touched

Diff

commit feef5fa560a89d73f984cdcf78eda8fb4aeeefab
Author: Trevin Chow <trevin@trevinchow.com>
Date:   Sat Mar 28 00:00:01 2026 -0700

    fix(skill): force agent readiness review to run in foreground
    
    The cli-agent-readiness-reviewer dispatch in Phase 4.9 was being
    backgrounded, causing the skill to skip straight to Phase 5 without
    waiting for results. Add explicit foreground-only instructions and
    run_in_background: false to both the initial dispatch and the re-run.
    
    Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---
 skills/printing-press/SKILL.md | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/skills/printing-press/SKILL.md b/skills/printing-press/SKILL.md
index ddb7a3fa..7a022f4c 100644
--- a/skills/printing-press/SKILL.md
+++ b/skills/printing-press/SKILL.md
@@ -1889,17 +1889,20 @@ Tell the user: "Runtime verification: [X]% pass rate ([N]/[M] commands). Data pi
 
 The existing agent-native scorecard dimension checks for flags. This phase goes deeper — evaluating 7 principles (non-interactive automation, structured output, progressive help, actionable errors, safe retries, composability, bounded responses) with file-level fix recommendations.
 
-### Step 4.9a: Dispatch the Agent (MANDATORY)
+### Step 4.9a: Dispatch the Agent (MANDATORY — FOREGROUND ONLY)
 
-You MUST execute this Agent tool call:
+You MUST execute this Agent tool call **in the foreground** (NOT background). You need the results before you can proceed — do NOT use `run_in_background: true`.
 
 ```
 Agent tool:
   subagent_type: compound-engineering:review:cli-agent-readiness-reviewer
+  run_in_background: false
   prompt: "Run the compound-engineering:cli-agent-readiness-reviewer agent on the <api> CLI in <output-dir>.
            Do not look at code elsewhere in the repo outside of that folder."
 ```
 
+**WAIT for the agent to return results before moving on.** Do NOT proceed to Phase 5 while this is running.
+
 **If the dispatch succeeds:** proceed to Step 4.9b with the results.
 
 **If the dispatch fails** (agent not found, plugin not installed, tool rejected):
@@ -1954,7 +1957,7 @@ All listed fixes are attempted — the reviewer already ranks by impact.
 
 ### Step 4.9d: Termination Check
 
-After implementing fixes, re-run the `compound-engineering:cli-agent-readiness-reviewer` agent on the same folder (same prompt as Step 4.9a). Evaluate the new scorecard:
+After implementing fixes, re-run the `compound-engineering:cli-agent-readiness-reviewer` agent on the same folder (same foreground dispatch as Step 4.9a — do NOT background). Evaluate the new scorecard:
 
 - **Zero Blockers AND zero Frictions:** Pass. Proceed to Phase 5.
 - **Blockers or Frictions remain, pass count < 2:** Return to Step 4.9c with the new fix list.

← cf3fcd99 docs(scorecard): add scoring architecture best-practice guid  ·  back to Cli Printing Press  ·  fix(skill): preserve codex fix delegation 0dca872a →