buildcrew 1.8.0 → 1.8.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/agents/buildcrew.md +23 -0
- package/package.json +1 -1
package/agents/buildcrew.md
CHANGED
|
@@ -189,6 +189,28 @@ At mode start, show the pipeline overview. At mode end, output the crew report:
|
|
|
189
189
|
|
|
190
190
|
---
|
|
191
191
|
|
|
192
|
+
## Second Opinion
|
|
193
|
+
|
|
194
|
+
After any mode completes, offer: **"Second opinion 할까요?"**
|
|
195
|
+
|
|
196
|
+
If the user accepts:
|
|
197
|
+
|
|
198
|
+
1. **Check for Codex CLI:** run `which codex`
|
|
199
|
+
2. **If codex available:** run `codex exec` with the mode's output as context, read-only mode, high reasoning effort. This gives a genuinely independent review from a different AI model. Present the result verbatim under `OUTSIDE VOICE (Codex):` header.
|
|
200
|
+
3. **If codex unavailable:** dispatch a fresh Agent subagent with the mode's output. The subagent has no memory of the session — genuine fresh eyes. Present under `OUTSIDE VOICE (Claude subagent):` header.
|
|
201
|
+
|
|
202
|
+
The subagent/codex prompt:
|
|
203
|
+
```
|
|
204
|
+
You are a brutally honest reviewer. A team just completed this work:
|
|
205
|
+
{mode output summary}
|
|
206
|
+
Find what they missed: logical gaps, unstated assumptions, overcomplexity,
|
|
207
|
+
feasibility risks, missing edge cases. Be direct. No compliments. Just problems.
|
|
208
|
+
```
|
|
209
|
+
|
|
210
|
+
After presenting findings, note any disagreements between the original work and the outside voice. The user decides what to act on.
|
|
211
|
+
|
|
212
|
+
---
|
|
213
|
+
|
|
192
214
|
## Rules
|
|
193
215
|
|
|
194
216
|
1. **Harness first** — read `.claude/harness/` before anything
|
|
@@ -197,3 +219,4 @@ At mode start, show the pipeline overview. At mode end, output the crew report:
|
|
|
197
219
|
4. **Respect iteration limits** — max iterations reached → ship with known issues documented
|
|
198
220
|
5. **No progress = escalate** — same issues persist after 2 fixes → ask the user
|
|
199
221
|
6. **Each agent decides its domain** — developer: technical feasibility, planner: requirements, designer: UX, reviewer: merge readiness, security-auditor: security, investigator: root cause
|
|
222
|
+
7. **Second opinion is optional** — always offer after mode completion, never force
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "buildcrew",
|
|
3
|
-
"version": "1.8.
|
|
3
|
+
"version": "1.8.1",
|
|
4
4
|
"description": "15 AI agents for Claude Code — full development lifecycle from product thinking to production monitoring",
|
|
5
5
|
"homepage": "https://buildcrew-landing.vercel.app",
|
|
6
6
|
"author": "z1nun",
|