specrails-core 3.0.0 → 3.1.0
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/commands/setup.md +1 -8
- package/install.sh +43 -6
- package/package.json +5 -1
- package/templates/commands/sr/team-debug.md +286 -0
- package/templates/commands/sr/team-review.md +341 -0
- package/update.sh +3 -3
package/commands/setup.md
CHANGED
|
@@ -314,21 +314,14 @@ When no flags are passed (the default), run this streamlined 3-question setup. D
|
|
|
314
314
|
|
|
315
315
|
### QS1: Ask the 3 questions
|
|
316
316
|
|
|
317
|
-
|
|
317
|
+
Display the following prompt EXACTLY ONCE and then wait for the user's responses. Do NOT repeat the questions — output them a single time only.
|
|
318
318
|
|
|
319
|
-
```
|
|
320
319
|
Welcome to specrails! Let's get your AI agent team set up in 3 quick questions.
|
|
321
320
|
|
|
322
321
|
1. What is this project? (one sentence)
|
|
323
|
-
> _
|
|
324
|
-
|
|
325
322
|
2. Who are the target users?
|
|
326
|
-
> _
|
|
327
|
-
|
|
328
323
|
3. Git access for agents — read-only or read-write?
|
|
329
324
|
(read-only = agents can read and suggest; read-write = agents can commit)
|
|
330
|
-
> _
|
|
331
|
-
```
|
|
332
325
|
|
|
333
326
|
Store the answers as:
|
|
334
327
|
- `QS_PROJECT_DESCRIPTION` — answer to question 1
|
package/install.sh
CHANGED
|
@@ -44,6 +44,7 @@ SPECRAILS_DIR=""
|
|
|
44
44
|
INSTRUCTIONS_FILE=""
|
|
45
45
|
HAS_CLAUDE=false
|
|
46
46
|
HAS_CODEX=false
|
|
47
|
+
AGENT_TEAMS=false
|
|
47
48
|
|
|
48
49
|
while [[ $# -gt 0 ]]; do
|
|
49
50
|
case "$1" in
|
|
@@ -98,7 +99,7 @@ if [[ -z "$CUSTOM_ROOT_DIR" && -f "$SCRIPT_DIR/install.sh" && -d "$SCRIPT_DIR/te
|
|
|
98
99
|
echo -e "${YELLOW}⚠${NC} You're running the installer from inside the specrails source repo."
|
|
99
100
|
echo -e " specrails installs into a ${BOLD}target${NC} repository, not into itself."
|
|
100
101
|
echo ""
|
|
101
|
-
read -p " Enter the path to the target repo (or 'q' to quit): " TARGET_PATH
|
|
102
|
+
read -p " Enter the path to the target repo (or 'q' to quit): " TARGET_PATH || TARGET_PATH="q"
|
|
102
103
|
if [[ "$TARGET_PATH" == "q" || -z "$TARGET_PATH" ]]; then
|
|
103
104
|
echo " Aborted. No changes made."
|
|
104
105
|
exit 0
|
|
@@ -229,7 +230,7 @@ elif [ "$HAS_CLAUDE" = true ] && [ "$HAS_CODEX" = true ]; then
|
|
|
229
230
|
echo " 1) Claude Code (claude) → output to .claude/"
|
|
230
231
|
echo " 2) Codex (codex) → output to .codex/"
|
|
231
232
|
echo ""
|
|
232
|
-
read -p " Select provider (1 or 2, default: 1): " PROVIDER_CHOICE
|
|
233
|
+
read -p " Select provider (1 or 2, default: 1): " PROVIDER_CHOICE || PROVIDER_CHOICE="1"
|
|
233
234
|
PROVIDER_CHOICE="${PROVIDER_CHOICE:-1}"
|
|
234
235
|
if [[ "$PROVIDER_CHOICE" == "2" ]]; then
|
|
235
236
|
CLI_PROVIDER="codex"
|
|
@@ -266,6 +267,35 @@ else
|
|
|
266
267
|
INSTRUCTIONS_FILE="CLAUDE.md"
|
|
267
268
|
fi
|
|
268
269
|
|
|
270
|
+
# 1.2b Agent Teams opt-in (Claude Code only)
|
|
271
|
+
if [[ "$CLI_PROVIDER" == "claude" ]]; then
|
|
272
|
+
echo ""
|
|
273
|
+
if [ "$AUTO_YES" = true ]; then
|
|
274
|
+
# --yes flag: default to NOT installing (opt-in, not opt-out)
|
|
275
|
+
AGENT_TEAMS=false
|
|
276
|
+
info "Agent Teams commands: skipped (opt-in, use interactive mode to enable)"
|
|
277
|
+
else
|
|
278
|
+
echo -e " ${BOLD}Agent Teams commands are available (experimental):${NC}"
|
|
279
|
+
echo " /sr:team-review — Multi-perspective code review with AI reviewers"
|
|
280
|
+
echo " /sr:team-debug — Collaborative debugging with competing hypotheses"
|
|
281
|
+
echo ""
|
|
282
|
+
echo " These require Claude Code Agent Teams (experimental feature)."
|
|
283
|
+
read -p " Install Agent Teams commands? (y/n, default: n): " INSTALL_AGENT_TEAMS || INSTALL_AGENT_TEAMS="n"
|
|
284
|
+
if [[ "$INSTALL_AGENT_TEAMS" == "y" || "$INSTALL_AGENT_TEAMS" == "Y" ]]; then
|
|
285
|
+
AGENT_TEAMS=true
|
|
286
|
+
ok "Agent Teams commands will be installed"
|
|
287
|
+
echo ""
|
|
288
|
+
info "Remember to set the feature flag before using these commands:"
|
|
289
|
+
echo ""
|
|
290
|
+
echo " export CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1"
|
|
291
|
+
echo ""
|
|
292
|
+
else
|
|
293
|
+
AGENT_TEAMS=false
|
|
294
|
+
info "Agent Teams commands: skipped"
|
|
295
|
+
fi
|
|
296
|
+
fi
|
|
297
|
+
fi
|
|
298
|
+
|
|
269
299
|
# 1.3 API key / authentication (provider-specific)
|
|
270
300
|
if [[ "$CLI_PROVIDER" == "claude" ]]; then
|
|
271
301
|
CLAUDE_AUTHED=false
|
|
@@ -334,7 +364,7 @@ elif [ -f "$REPO_ROOT/node_modules/.bin/openspec" ]; then
|
|
|
334
364
|
else
|
|
335
365
|
warn "OpenSpec CLI not found."
|
|
336
366
|
if [ "$HAS_NPM" = true ]; then
|
|
337
|
-
if [ "$AUTO_YES" = true ]; then INSTALL_OPENSPEC="y"; else read -p " Install OpenSpec CLI globally? (y/n): " INSTALL_OPENSPEC; fi
|
|
367
|
+
if [ "$AUTO_YES" = true ]; then INSTALL_OPENSPEC="y"; else read -p " Install OpenSpec CLI globally? (y/n): " INSTALL_OPENSPEC || INSTALL_OPENSPEC="n"; fi
|
|
338
368
|
if [ "$INSTALL_OPENSPEC" = "y" ] || [ "$INSTALL_OPENSPEC" = "Y" ]; then
|
|
339
369
|
info "Installing OpenSpec CLI..."
|
|
340
370
|
npm install -g @openspec/cli 2>/dev/null && {
|
|
@@ -439,7 +469,7 @@ fi
|
|
|
439
469
|
if [ "$EXISTING_SETUP" = true ]; then
|
|
440
470
|
echo ""
|
|
441
471
|
warn "This repo already has some agent/command/openspec artifacts."
|
|
442
|
-
if [ "$AUTO_YES" = true ]; then CONTINUE="y"; else read -p " Continue and merge with existing setup? (y/n): " CONTINUE; fi
|
|
472
|
+
if [ "$AUTO_YES" = true ]; then CONTINUE="y"; else read -p " Continue and merge with existing setup? (y/n): " CONTINUE || CONTINUE="n"; fi
|
|
443
473
|
if [ "$CONTINUE" != "y" ] && [ "$CONTINUE" != "Y" ]; then
|
|
444
474
|
info "Aborted. No changes made."
|
|
445
475
|
exit 0
|
|
@@ -521,7 +551,9 @@ chmod +x "$REPO_ROOT/.specrails/bin/doctor.sh"
|
|
|
521
551
|
ok "Installed specrails doctor (bin/doctor.sh)"
|
|
522
552
|
|
|
523
553
|
# Copy templates (includes commands, skills, agents, rules, personas, settings)
|
|
524
|
-
|
|
554
|
+
# Use tar to exclude node_modules and package-lock.json for performance
|
|
555
|
+
tar -C "$SCRIPT_DIR/templates" --exclude='node_modules' --exclude='package-lock.json' -cf - . \
|
|
556
|
+
| tar -C "$REPO_ROOT/$SPECRAILS_DIR/setup-templates/" -xf -
|
|
525
557
|
ok "Installed setup templates (commands + skills)"
|
|
526
558
|
|
|
527
559
|
# Write OSS detection results for /setup
|
|
@@ -542,7 +574,8 @@ cat > "$REPO_ROOT/$SPECRAILS_DIR/setup-templates/.provider-detection.json" << EO
|
|
|
542
574
|
{
|
|
543
575
|
"cli_provider": "$CLI_PROVIDER",
|
|
544
576
|
"specrails_dir": "$SPECRAILS_DIR",
|
|
545
|
-
"instructions_file": "$INSTRUCTIONS_FILE"
|
|
577
|
+
"instructions_file": "$INSTRUCTIONS_FILE",
|
|
578
|
+
"agent_teams": $AGENT_TEAMS
|
|
546
579
|
}
|
|
547
580
|
EOF
|
|
548
581
|
ok "Provider detection results written ($CLI_PROVIDER → $SPECRAILS_DIR/)"
|
|
@@ -589,6 +622,10 @@ echo ""
|
|
|
589
622
|
echo -e "${BOLD}${GREEN}Installation summary:${NC}"
|
|
590
623
|
echo ""
|
|
591
624
|
echo " Provider: $CLI_PROVIDER → output to $SPECRAILS_DIR/"
|
|
625
|
+
if [[ "$AGENT_TEAMS" == "true" ]]; then
|
|
626
|
+
echo " Agent Teams: installed (/sr:team-review, /sr:team-debug)"
|
|
627
|
+
echo " Feature flag: CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1 (required)"
|
|
628
|
+
fi
|
|
592
629
|
echo ""
|
|
593
630
|
echo " Files installed:"
|
|
594
631
|
if [[ "$CLI_PROVIDER" == "codex" ]]; then
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "specrails-core",
|
|
3
|
-
"version": "3.
|
|
3
|
+
"version": "3.1.0",
|
|
4
4
|
"description": "AI agent workflow system for Claude Code — installs 12 specialized agents, orchestration commands, and persona-driven product discovery into any repository",
|
|
5
5
|
"bin": {
|
|
6
6
|
"specrails-core": "bin/specrails-core.js"
|
|
@@ -50,6 +50,10 @@
|
|
|
50
50
|
"bugs": {
|
|
51
51
|
"url": "https://github.com/fjpulidop/specrails-core/issues"
|
|
52
52
|
},
|
|
53
|
+
"scripts": {
|
|
54
|
+
"test": "bash tests/run-all.sh",
|
|
55
|
+
"test:coverage": "bash tests/run-coverage.sh"
|
|
56
|
+
},
|
|
53
57
|
"license": "MIT",
|
|
54
58
|
"author": "fjpulidop"
|
|
55
59
|
}
|
|
@@ -0,0 +1,286 @@
|
|
|
1
|
+
---
|
|
2
|
+
name: "Team Debug"
|
|
3
|
+
description: "Collaborative debugging using Claude Code Agent Teams. Multiple investigators pursue competing hypotheses independently, challenge each other's findings, and produce a root cause analysis ranked by confidence."
|
|
4
|
+
category: Workflow
|
|
5
|
+
tags: [workflow, debugging, agent-teams, investigation]
|
|
6
|
+
---
|
|
7
|
+
|
|
8
|
+
Collaborative debugging for **{{PROJECT_NAME}}** using Claude Code Agent Teams. Multiple investigators pursue competing hypotheses about a bug, challenge each other's findings, and produce a confidence-ranked root cause analysis.
|
|
9
|
+
|
|
10
|
+
**Input:** $ARGUMENTS — required: a bug description in one of these forms:
|
|
11
|
+
- `"Login fails silently when email has uppercase letters"` — free text bug description
|
|
12
|
+
- `tests/auth.test.ts` — a failing test file path
|
|
13
|
+
- `"TypeError: Cannot read property 'id' of undefined"` — an error message or stack trace
|
|
14
|
+
- `#42` — a GitHub issue number
|
|
15
|
+
|
|
16
|
+
Optional flags:
|
|
17
|
+
- `--scope <paths>` — comma-separated file/directory paths to constrain investigation (default: entire repo)
|
|
18
|
+
- `--depth <level>` — investigation depth: `shallow`, `normal` (default), `deep`
|
|
19
|
+
|
|
20
|
+
---
|
|
21
|
+
|
|
22
|
+
## Phase 0: Feature Flag Guard
|
|
23
|
+
|
|
24
|
+
**This check is mandatory and runs before anything else.**
|
|
25
|
+
|
|
26
|
+
Check whether Agent Teams is enabled:
|
|
27
|
+
|
|
28
|
+
```bash
|
|
29
|
+
echo "${CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS:-}"
|
|
30
|
+
```
|
|
31
|
+
|
|
32
|
+
If the variable is unset or not equal to `1`, print this message and **stop immediately**:
|
|
33
|
+
|
|
34
|
+
```
|
|
35
|
+
Error: Agent Teams is an experimental feature that requires opt-in.
|
|
36
|
+
|
|
37
|
+
To enable it, set the environment variable before starting Claude Code:
|
|
38
|
+
|
|
39
|
+
export CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1
|
|
40
|
+
|
|
41
|
+
Agent Teams requires Claude Code v2.1.32 or later.
|
|
42
|
+
```
|
|
43
|
+
|
|
44
|
+
Do NOT proceed past this point if the guard fails.
|
|
45
|
+
|
|
46
|
+
---
|
|
47
|
+
|
|
48
|
+
## Phase 1: Bug Context Gathering
|
|
49
|
+
|
|
50
|
+
Parse `$ARGUMENTS` to determine the bug context and flags.
|
|
51
|
+
|
|
52
|
+
**Variables to set:**
|
|
53
|
+
|
|
54
|
+
- `BUG_DESCRIPTION` — the user's description of the bug
|
|
55
|
+
- `INPUT_TYPE` — `"text"`, `"test"`, `"error"`, or `"issue"`
|
|
56
|
+
- `SCOPE_PATHS` — from `--scope` flag or `["."]` (entire repo)
|
|
57
|
+
- `INVESTIGATION_DEPTH` — from `--depth` flag or `"normal"`
|
|
58
|
+
|
|
59
|
+
**Detection rules:**
|
|
60
|
+
|
|
61
|
+
1. If input starts with `#` or is a bare integer -> `INPUT_TYPE="issue"`, fetch issue body via `gh issue view <number> --json title,body`
|
|
62
|
+
2. If input is a file path that exists and contains `test` or `spec` -> `INPUT_TYPE="test"`, run the test to capture failure output
|
|
63
|
+
3. If input contains common error patterns (stack trace, `Error:`, `Exception`, `FATAL`) -> `INPUT_TYPE="error"`
|
|
64
|
+
4. Otherwise -> `INPUT_TYPE="text"`
|
|
65
|
+
|
|
66
|
+
If `$ARGUMENTS` is empty, print usage and stop:
|
|
67
|
+
```
|
|
68
|
+
Usage: /sr:team-debug <bug-description> [--scope <paths>] [--depth <level>]
|
|
69
|
+
|
|
70
|
+
Examples:
|
|
71
|
+
/sr:team-debug "Login fails silently when email has uppercase letters"
|
|
72
|
+
/sr:team-debug tests/auth.test.ts --depth deep
|
|
73
|
+
/sr:team-debug "TypeError: Cannot read property 'id' of undefined" --scope src/api
|
|
74
|
+
/sr:team-debug #42
|
|
75
|
+
```
|
|
76
|
+
|
|
77
|
+
---
|
|
78
|
+
|
|
79
|
+
## Phase 2: Hypothesis Generation
|
|
80
|
+
|
|
81
|
+
Analyze the bug context and generate competing hypotheses:
|
|
82
|
+
|
|
83
|
+
1. **Analyze symptoms**: Read relevant source files within `SCOPE_PATHS`, examine recent git changes (`git log --oneline -20` for the relevant paths), check related tests
|
|
84
|
+
2. **Generate hypotheses**: Produce 2-3 distinct, testable hypotheses about the root cause. Each must target a different subsystem, mechanism, or failure mode.
|
|
85
|
+
3. **Rank hypotheses**: Order by initial likelihood based on available evidence
|
|
86
|
+
|
|
87
|
+
Each hypothesis MUST include:
|
|
88
|
+
- **Title**: One-line description of the proposed cause
|
|
89
|
+
- **Rationale**: Why this is plausible given the symptoms
|
|
90
|
+
- **Investigation plan**: Specific files to examine, commands to run, patterns to search for
|
|
91
|
+
- **Expected evidence**: What would confirm or refute this hypothesis
|
|
92
|
+
|
|
93
|
+
Decide whether to launch 2 or 3 investigators:
|
|
94
|
+
- Default: 2 investigators (Hypothesis A and B)
|
|
95
|
+
- Launch 3 when: `--depth deep` is set, OR the bug description is ambiguous enough to warrant three genuinely distinct hypotheses
|
|
96
|
+
|
|
97
|
+
Print the hypothesis summary:
|
|
98
|
+
```
|
|
99
|
+
## Bug Analysis
|
|
100
|
+
|
|
101
|
+
**Input type:** <text / test / error / issue>
|
|
102
|
+
**Scope:** <SCOPE_PATHS>
|
|
103
|
+
**Depth:** <INVESTIGATION_DEPTH>
|
|
104
|
+
|
|
105
|
+
### Hypotheses
|
|
106
|
+
|
|
107
|
+
1. **<Hypothesis A title>** (initial confidence: X%)
|
|
108
|
+
<one-line rationale>
|
|
109
|
+
|
|
110
|
+
2. **<Hypothesis B title>** (initial confidence: Y%)
|
|
111
|
+
<one-line rationale>
|
|
112
|
+
|
|
113
|
+
3. **<Hypothesis C title>** (initial confidence: Z%) [if applicable]
|
|
114
|
+
<one-line rationale>
|
|
115
|
+
|
|
116
|
+
Launching <N> investigators...
|
|
117
|
+
```
|
|
118
|
+
|
|
119
|
+
---
|
|
120
|
+
|
|
121
|
+
## Phase 3: Launch Investigation Team
|
|
122
|
+
|
|
123
|
+
Create 2-3 investigator teammates using Agent Teams. Each teammate receives the bug context and their assigned hypothesis.
|
|
124
|
+
|
|
125
|
+
**IMPORTANT:** Use the Agent Teams teammate mechanism — NOT the Agent tool's `subagent_type`. Teammates share a task list and can message each other via mailbox.
|
|
126
|
+
|
|
127
|
+
### Investigator A: Primary Hypothesis
|
|
128
|
+
|
|
129
|
+
**Prompt:**
|
|
130
|
+
```
|
|
131
|
+
You are Investigator A on a debugging team. Your job is to prove or disprove your assigned hypothesis through systematic investigation.
|
|
132
|
+
|
|
133
|
+
## Bug Context
|
|
134
|
+
<BUG_DESCRIPTION>
|
|
135
|
+
<additional context gathered in Phase 1: test output, error details, issue body>
|
|
136
|
+
|
|
137
|
+
## Your Hypothesis
|
|
138
|
+
**<Hypothesis A title>**
|
|
139
|
+
<Hypothesis A rationale>
|
|
140
|
+
|
|
141
|
+
## Investigation Plan
|
|
142
|
+
<Hypothesis A investigation plan>
|
|
143
|
+
|
|
144
|
+
## Scope
|
|
145
|
+
Limit your investigation to: <SCOPE_PATHS>
|
|
146
|
+
|
|
147
|
+
## Instructions
|
|
148
|
+
1. Follow your investigation plan systematically
|
|
149
|
+
2. Read source files, search for patterns, examine git history within scope
|
|
150
|
+
3. Collect concrete evidence — file paths, line numbers, code snippets, command output
|
|
151
|
+
4. Determine a confidence level (0-100%) that your hypothesis is the root cause
|
|
152
|
+
5. After completing your investigation, read the other investigators' findings from the task list
|
|
153
|
+
6. If you find evidence that supports or contradicts another investigator's hypothesis, send a mailbox message challenging or endorsing their findings
|
|
154
|
+
7. Do NOT modify any files — this is a read-only investigation
|
|
155
|
+
|
|
156
|
+
## Report Format
|
|
157
|
+
Post your findings as a task list update:
|
|
158
|
+
|
|
159
|
+
### Investigator A: <Hypothesis Title>
|
|
160
|
+
|
|
161
|
+
#### Evidence Found
|
|
162
|
+
| # | Type | Location | Finding |
|
|
163
|
+
|---|------|----------|---------|
|
|
164
|
+
| 1 | code/git/test/config | file:line | What was found |
|
|
165
|
+
|
|
166
|
+
#### Evidence Against
|
|
167
|
+
<any evidence that contradicts this hypothesis>
|
|
168
|
+
|
|
169
|
+
#### Confidence: X%
|
|
170
|
+
<explanation of confidence level>
|
|
171
|
+
|
|
172
|
+
#### Verdict
|
|
173
|
+
<CONFIRMED / REFUTED / INCONCLUSIVE>
|
|
174
|
+
<brief explanation>
|
|
175
|
+
```
|
|
176
|
+
|
|
177
|
+
### Investigator B: Alternative Hypothesis
|
|
178
|
+
|
|
179
|
+
Same structure as Investigator A, with Hypothesis B title, rationale, investigation plan, and expected evidence.
|
|
180
|
+
|
|
181
|
+
### Investigator C: Contrarian Hypothesis (when applicable)
|
|
182
|
+
|
|
183
|
+
Same structure as Investigator A, with Hypothesis C title, rationale, investigation plan, and expected evidence. Only launched when 3 hypotheses are generated.
|
|
184
|
+
|
|
185
|
+
### Team Coordination
|
|
186
|
+
|
|
187
|
+
After launching all investigators:
|
|
188
|
+
|
|
189
|
+
1. Wait for all investigators to complete their independent investigations (posted to the shared task list)
|
|
190
|
+
2. Allow one round of cross-investigation challenge via mailbox — each investigator may respond to findings from the others
|
|
191
|
+
3. Collect all findings and challenge outcomes
|
|
192
|
+
|
|
193
|
+
If any teammate fails to respond, proceed with available findings and note the gap in the final report.
|
|
194
|
+
|
|
195
|
+
---
|
|
196
|
+
|
|
197
|
+
## Phase 4: Root Cause Synthesis
|
|
198
|
+
|
|
199
|
+
After all investigations and challenges are complete, the team lead produces the root cause analysis.
|
|
200
|
+
|
|
201
|
+
### Step 1: Evidence Aggregation
|
|
202
|
+
|
|
203
|
+
1. Read all investigator reports from the task list
|
|
204
|
+
2. Identify converging evidence (multiple investigators pointing to the same cause)
|
|
205
|
+
3. Identify conflicting evidence and note unresolved disagreements
|
|
206
|
+
4. Read any mailbox challenge messages for cross-investigator insights
|
|
207
|
+
|
|
208
|
+
### Step 2: Confidence Recalibration
|
|
209
|
+
|
|
210
|
+
Adjust hypothesis confidence levels based on cross-investigation challenges:
|
|
211
|
+
- A hypothesis endorsed by evidence from another investigator: confidence boost (+10-20%)
|
|
212
|
+
- A hypothesis contradicted by another investigator's evidence: confidence reduction (-10-30%)
|
|
213
|
+
- A hypothesis with no cross-investigation interaction: confidence unchanged
|
|
214
|
+
- Converging evidence from multiple investigators: strongest signal for root cause
|
|
215
|
+
|
|
216
|
+
### Step 3: Render Report
|
|
217
|
+
|
|
218
|
+
```markdown
|
|
219
|
+
## Root Cause Analysis
|
|
220
|
+
|
|
221
|
+
**Bug:** <one-line bug description>
|
|
222
|
+
**Scope:** <SCOPE_PATHS>
|
|
223
|
+
**Depth:** <INVESTIGATION_DEPTH>
|
|
224
|
+
**Investigators:** <N>
|
|
225
|
+
|
|
226
|
+
---
|
|
227
|
+
|
|
228
|
+
### Most Likely Root Cause
|
|
229
|
+
|
|
230
|
+
**<Winning hypothesis title>** — Confidence: X%
|
|
231
|
+
|
|
232
|
+
<2-3 sentence explanation of the root cause with specific file and line references>
|
|
233
|
+
|
|
234
|
+
#### Key Evidence
|
|
235
|
+
| # | Location | Finding |
|
|
236
|
+
|---|----------|---------|
|
|
237
|
+
| 1 | file:line | Evidence supporting this conclusion |
|
|
238
|
+
|
|
239
|
+
#### Suggested Fix
|
|
240
|
+
<specific, actionable fix recommendation with file and line references>
|
|
241
|
+
|
|
242
|
+
---
|
|
243
|
+
|
|
244
|
+
### Alternative Hypotheses
|
|
245
|
+
|
|
246
|
+
| # | Hypothesis | Initial | Final | Verdict | Key Evidence |
|
|
247
|
+
|---|-----------|---------|-------|---------|-------------|
|
|
248
|
+
| 1 | <title> | X% | Y% | CONFIRMED/REFUTED/INCONCLUSIVE | <one-line> |
|
|
249
|
+
| 2 | <title> | X% | Y% | CONFIRMED/REFUTED/INCONCLUSIVE | <one-line> |
|
|
250
|
+
|
|
251
|
+
### Investigation Trail
|
|
252
|
+
|
|
253
|
+
<for each investigator, a 2-3 line summary of what they investigated and what they found>
|
|
254
|
+
|
|
255
|
+
### Cross-Investigation Notes
|
|
256
|
+
|
|
257
|
+
<any points of challenge or agreement between investigators, with resolution>
|
|
258
|
+
|
|
259
|
+
---
|
|
260
|
+
|
|
261
|
+
### Recommended Next Steps
|
|
262
|
+
|
|
263
|
+
1. <most important action — typically the fix>
|
|
264
|
+
2. <verification step — test to confirm the fix>
|
|
265
|
+
3. <preventive step — how to avoid this class of bug>
|
|
266
|
+
```
|
|
267
|
+
|
|
268
|
+
### Step 4: Cost Notice
|
|
269
|
+
|
|
270
|
+
Print a brief cost notice after the report:
|
|
271
|
+
|
|
272
|
+
```
|
|
273
|
+
Note: Team debug used ~<N>x the tokens of a single-agent investigation (<N> parallel investigators + challenge round).
|
|
274
|
+
```
|
|
275
|
+
|
|
276
|
+
---
|
|
277
|
+
|
|
278
|
+
## Rules
|
|
279
|
+
|
|
280
|
+
- This command is **read-only** — it MUST NOT create, modify, or delete any files
|
|
281
|
+
- All investigators run as Agent Teams teammates, not as Agent tool subagents
|
|
282
|
+
- If Agent Teams is unavailable at runtime (API error, version mismatch), fall back to running 2-3 sequential Agent tool subagents and skip the challenge phase. Print a warning about the fallback.
|
|
283
|
+
- The challenge phase is limited to one round per investigator to control token costs
|
|
284
|
+
- Findings MUST include file paths and line numbers — vague findings are not acceptable
|
|
285
|
+
- Investigation MUST stay within `SCOPE_PATHS` if specified
|
|
286
|
+
- This command is Claude Code only — it is NOT installed when the provider is `codex`
|
|
@@ -0,0 +1,341 @@
|
|
|
1
|
+
---
|
|
2
|
+
name: "Team Review"
|
|
3
|
+
description: "Multi-perspective code review using Claude Code Agent Teams. Three specialized reviewers (security, performance, correctness) independently review changes, debate findings, and produce a consolidated report."
|
|
4
|
+
category: Workflow
|
|
5
|
+
tags: [workflow, review, agent-teams, security, performance]
|
|
6
|
+
---
|
|
7
|
+
|
|
8
|
+
Multi-perspective code review for **{{PROJECT_NAME}}** using Claude Code Agent Teams. Three specialized reviewers analyze changes independently, debate cross-cutting findings, and produce a consolidated report.
|
|
9
|
+
|
|
10
|
+
**Input:** $ARGUMENTS — required: a review target in one of these forms:
|
|
11
|
+
- `#123` — review a pull request by number
|
|
12
|
+
- `feat/my-feature` — review a branch diff against base
|
|
13
|
+
- `abc1234..def5678` — review a commit range
|
|
14
|
+
|
|
15
|
+
Optional flags:
|
|
16
|
+
- `--base <branch>` — override base branch for comparison (default: repository default branch)
|
|
17
|
+
- `--focus <areas>` — comma-separated focus areas to weight: `security`, `performance`, `correctness`, `tests`, `types`
|
|
18
|
+
|
|
19
|
+
---
|
|
20
|
+
|
|
21
|
+
## Phase 0: Feature Flag Guard
|
|
22
|
+
|
|
23
|
+
**This check is mandatory and runs before anything else.**
|
|
24
|
+
|
|
25
|
+
Check whether Agent Teams is enabled:
|
|
26
|
+
|
|
27
|
+
```bash
|
|
28
|
+
echo "${CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS:-}"
|
|
29
|
+
```
|
|
30
|
+
|
|
31
|
+
If the variable is unset or not equal to `1`, print this message and **stop immediately**:
|
|
32
|
+
|
|
33
|
+
```
|
|
34
|
+
Error: Agent Teams is an experimental feature that requires opt-in.
|
|
35
|
+
|
|
36
|
+
To enable it, set the environment variable before starting Claude Code:
|
|
37
|
+
|
|
38
|
+
export CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1
|
|
39
|
+
|
|
40
|
+
Agent Teams requires Claude Code v2.1.32 or later.
|
|
41
|
+
```
|
|
42
|
+
|
|
43
|
+
Do NOT proceed past this point if the guard fails.
|
|
44
|
+
|
|
45
|
+
---
|
|
46
|
+
|
|
47
|
+
## Phase 1: Input Parsing
|
|
48
|
+
|
|
49
|
+
Parse `$ARGUMENTS` to determine the review target and flags.
|
|
50
|
+
|
|
51
|
+
**Variables to set:**
|
|
52
|
+
|
|
53
|
+
- `REVIEW_TARGET` — the PR number, branch name, or commit range
|
|
54
|
+
- `REVIEW_TYPE` — `"pr"`, `"branch"`, or `"range"`
|
|
55
|
+
- `BASE_BRANCH` — from `--base` flag or detect via `gh repo view --json defaultBranchRef -q '.defaultBranchRef.name'` or fall back to `main`
|
|
56
|
+
- `FOCUS_AREAS` — array from `--focus` flag or `["all"]`
|
|
57
|
+
|
|
58
|
+
**Detection rules:**
|
|
59
|
+
|
|
60
|
+
1. If input starts with `#` or is a bare integer → `REVIEW_TYPE="pr"`, strip `#` prefix
|
|
61
|
+
2. If input contains `..` → `REVIEW_TYPE="range"`
|
|
62
|
+
3. Otherwise → `REVIEW_TYPE="branch"`
|
|
63
|
+
|
|
64
|
+
If `$ARGUMENTS` is empty, print usage and stop:
|
|
65
|
+
```
|
|
66
|
+
Usage: /sr:team-review <target> [--base <branch>] [--focus <areas>]
|
|
67
|
+
|
|
68
|
+
Examples:
|
|
69
|
+
/sr:team-review #42
|
|
70
|
+
/sr:team-review feat/new-auth --focus security
|
|
71
|
+
/sr:team-review abc123..def456
|
|
72
|
+
```
|
|
73
|
+
|
|
74
|
+
---
|
|
75
|
+
|
|
76
|
+
## Phase 2: Gather Diff
|
|
77
|
+
|
|
78
|
+
Collect the code changes based on `REVIEW_TYPE`:
|
|
79
|
+
|
|
80
|
+
- **PR**: Run `gh pr diff <REVIEW_TARGET>` and `gh pr diff <REVIEW_TARGET> --name-only`
|
|
81
|
+
- **Branch**: Run `git diff ${BASE_BRANCH}...${REVIEW_TARGET}` and `git diff --name-only ${BASE_BRANCH}...${REVIEW_TARGET}`
|
|
82
|
+
- **Range**: Run `git diff ${REVIEW_TARGET}` and `git diff --name-only ${REVIEW_TARGET}`
|
|
83
|
+
|
|
84
|
+
Also collect file-level stats: `git diff --stat <appropriate-range>`
|
|
85
|
+
|
|
86
|
+
**Store these variables for Phase 3:**
|
|
87
|
+
- `DIFF_CONTENT` — full unified diff
|
|
88
|
+
- `CHANGED_FILES` — list of changed file paths
|
|
89
|
+
- `DIFF_STATS` — file-level line count changes
|
|
90
|
+
|
|
91
|
+
If the diff is empty, print `No changes found for the given review target.` and stop.
|
|
92
|
+
|
|
93
|
+
Print a summary:
|
|
94
|
+
```
|
|
95
|
+
## Review Target
|
|
96
|
+
Type: <PR / Branch / Range>
|
|
97
|
+
Target: <REVIEW_TARGET>
|
|
98
|
+
Base: <BASE_BRANCH>
|
|
99
|
+
Changed files: <N>
|
|
100
|
+
Focus: <FOCUS_AREAS or "all areas">
|
|
101
|
+
|
|
102
|
+
<DIFF_STATS output>
|
|
103
|
+
```
|
|
104
|
+
|
|
105
|
+
---
|
|
106
|
+
|
|
107
|
+
## Phase 3: Launch Team Review
|
|
108
|
+
|
|
109
|
+
Create three reviewer teammates using Agent Teams. Each teammate receives the full diff and file list.
|
|
110
|
+
|
|
111
|
+
**IMPORTANT:** Use the Agent Teams teammate mechanism — NOT the Agent tool's `subagent_type`. Teammates share a task list and can message each other via mailbox.
|
|
112
|
+
|
|
113
|
+
### Teammate 1: Security Reviewer
|
|
114
|
+
|
|
115
|
+
**Persona:** sr-security-reviewer (or sr-reviewer with security focus if persona not available)
|
|
116
|
+
|
|
117
|
+
**Prompt:**
|
|
118
|
+
```
|
|
119
|
+
You are the Security Reviewer on a team code review.
|
|
120
|
+
|
|
121
|
+
## Your Focus Areas
|
|
122
|
+
- Authentication and authorization flaws
|
|
123
|
+
- Input validation and injection vulnerabilities (SQL, XSS, command injection)
|
|
124
|
+
- Secrets or credentials in code
|
|
125
|
+
- OWASP Top 10 vulnerabilities
|
|
126
|
+
- Insecure dependencies or configurations
|
|
127
|
+
- Missing rate limiting or access controls
|
|
128
|
+
|
|
129
|
+
## Changed Files
|
|
130
|
+
<CHANGED_FILES>
|
|
131
|
+
|
|
132
|
+
## Diff
|
|
133
|
+
<DIFF_CONTENT>
|
|
134
|
+
|
|
135
|
+
## Instructions
|
|
136
|
+
1. Review every changed file through a security lens
|
|
137
|
+
2. Report findings using the format below — be specific about file, line, and fix
|
|
138
|
+
3. After completing your review, read the other reviewers' findings from the task list
|
|
139
|
+
4. If you have security-relevant context on their findings, send a mailbox message
|
|
140
|
+
|
|
141
|
+
## Report Format
|
|
142
|
+
Post your findings as a task list update:
|
|
143
|
+
|
|
144
|
+
### Security Review Findings
|
|
145
|
+
|
|
146
|
+
#### Summary
|
|
147
|
+
<1-2 sentences>
|
|
148
|
+
|
|
149
|
+
#### Findings
|
|
150
|
+
| # | Severity | File | Line(s) | Finding | Recommendation |
|
|
151
|
+
|---|----------|------|---------|---------|----------------|
|
|
152
|
+
|
|
153
|
+
#### Verdict
|
|
154
|
+
<APPROVE / REQUEST_CHANGES / COMMENT>
|
|
155
|
+
```
|
|
156
|
+
|
|
157
|
+
### Teammate 2: Performance Reviewer
|
|
158
|
+
|
|
159
|
+
**Persona:** sr-performance-reviewer (or sr-reviewer with performance focus if persona not available)
|
|
160
|
+
|
|
161
|
+
**Prompt:**
|
|
162
|
+
```
|
|
163
|
+
You are the Performance Reviewer on a team code review.
|
|
164
|
+
|
|
165
|
+
## Your Focus Areas
|
|
166
|
+
- Database query complexity and N+1 queries
|
|
167
|
+
- Missing or broken caching
|
|
168
|
+
- Memory leaks and excessive allocations
|
|
169
|
+
- Algorithmic complexity (O(n^2) or worse in hot paths)
|
|
170
|
+
- Bundle size and lazy loading concerns (frontend)
|
|
171
|
+
- Missing pagination or unbounded data fetching
|
|
172
|
+
|
|
173
|
+
## Changed Files
|
|
174
|
+
<CHANGED_FILES>
|
|
175
|
+
|
|
176
|
+
## Diff
|
|
177
|
+
<DIFF_CONTENT>
|
|
178
|
+
|
|
179
|
+
## Instructions
|
|
180
|
+
1. Review every changed file through a performance lens
|
|
181
|
+
2. Report findings using the format below — be specific about file, line, and fix
|
|
182
|
+
3. After completing your review, read the other reviewers' findings from the task list
|
|
183
|
+
4. If you have performance-relevant context on their findings, send a mailbox message
|
|
184
|
+
|
|
185
|
+
## Report Format
|
|
186
|
+
Post your findings as a task list update:
|
|
187
|
+
|
|
188
|
+
### Performance Review Findings
|
|
189
|
+
|
|
190
|
+
#### Summary
|
|
191
|
+
<1-2 sentences>
|
|
192
|
+
|
|
193
|
+
#### Findings
|
|
194
|
+
| # | Severity | File | Line(s) | Finding | Recommendation |
|
|
195
|
+
|---|----------|------|---------|---------|----------------|
|
|
196
|
+
|
|
197
|
+
#### Verdict
|
|
198
|
+
<APPROVE / REQUEST_CHANGES / COMMENT>
|
|
199
|
+
```
|
|
200
|
+
|
|
201
|
+
### Teammate 3: Correctness Reviewer
|
|
202
|
+
|
|
203
|
+
**Persona:** sr-reviewer
|
|
204
|
+
|
|
205
|
+
**Prompt:**
|
|
206
|
+
```
|
|
207
|
+
You are the Correctness Reviewer on a team code review.
|
|
208
|
+
|
|
209
|
+
## Your Focus Areas
|
|
210
|
+
- Logic errors and edge cases
|
|
211
|
+
- Missing or inadequate test coverage
|
|
212
|
+
- Type safety violations
|
|
213
|
+
- Error handling gaps (uncaught exceptions, missing error paths)
|
|
214
|
+
- API contract mismatches
|
|
215
|
+
- Race conditions and concurrency issues
|
|
216
|
+
|
|
217
|
+
## Changed Files
|
|
218
|
+
<CHANGED_FILES>
|
|
219
|
+
|
|
220
|
+
## Diff
|
|
221
|
+
<DIFF_CONTENT>
|
|
222
|
+
|
|
223
|
+
## Instructions
|
|
224
|
+
1. Review every changed file through a correctness and test coverage lens
|
|
225
|
+
2. Report findings using the format below — be specific about file, line, and fix
|
|
226
|
+
3. After completing your review, read the other reviewers' findings from the task list
|
|
227
|
+
4. If you have correctness-relevant context on their findings, send a mailbox message
|
|
228
|
+
|
|
229
|
+
## Report Format
|
|
230
|
+
Post your findings as a task list update:
|
|
231
|
+
|
|
232
|
+
### Correctness Review Findings
|
|
233
|
+
|
|
234
|
+
#### Summary
|
|
235
|
+
<1-2 sentences>
|
|
236
|
+
|
|
237
|
+
#### Findings
|
|
238
|
+
| # | Severity | File | Line(s) | Finding | Recommendation |
|
|
239
|
+
|---|----------|------|---------|---------|----------------|
|
|
240
|
+
|
|
241
|
+
#### Verdict
|
|
242
|
+
<APPROVE / REQUEST_CHANGES / COMMENT>
|
|
243
|
+
```
|
|
244
|
+
|
|
245
|
+
### Team Coordination
|
|
246
|
+
|
|
247
|
+
After launching all three teammates:
|
|
248
|
+
|
|
249
|
+
1. Wait for all three to complete their independent reviews (posted to the shared task list)
|
|
250
|
+
2. Allow one round of cross-review debate via mailbox — each reviewer may respond to findings from the other two
|
|
251
|
+
3. Collect all findings and debate outcomes
|
|
252
|
+
|
|
253
|
+
If any teammate fails to respond, proceed with available reviews and note the gap in the final report.
|
|
254
|
+
|
|
255
|
+
---
|
|
256
|
+
|
|
257
|
+
## Phase 4: Synthesize Consolidated Report
|
|
258
|
+
|
|
259
|
+
After all reviews and debate are complete, the team lead produces the final report.
|
|
260
|
+
|
|
261
|
+
### Step 1: Collect and Deduplicate
|
|
262
|
+
|
|
263
|
+
1. Read all three reviewer reports from the task list
|
|
264
|
+
2. Identify duplicate findings (same file + overlapping lines + similar issue)
|
|
265
|
+
3. For duplicates: keep the highest-severity version, note which reviewers flagged it
|
|
266
|
+
|
|
267
|
+
### Step 2: Apply Focus Weighting
|
|
268
|
+
|
|
269
|
+
If `FOCUS_AREAS` is not `["all"]`:
|
|
270
|
+
- Findings in focus areas get their severity preserved
|
|
271
|
+
- Findings outside focus areas: Critical stays Critical, but High→Medium, Medium→Low for display purposes
|
|
272
|
+
- Note the weighting in the report header
|
|
273
|
+
|
|
274
|
+
### Step 3: Render Report
|
|
275
|
+
|
|
276
|
+
```markdown
|
|
277
|
+
## Team Review Report
|
|
278
|
+
|
|
279
|
+
**Target:** <PR #N / branch-name / commit-range>
|
|
280
|
+
**Base:** <BASE_BRANCH>
|
|
281
|
+
**Reviewers:** Security, Performance, Correctness
|
|
282
|
+
**Changed files:** N files (+X/-Y lines)
|
|
283
|
+
**Focus:** <FOCUS_AREAS or "all areas equally weighted">
|
|
284
|
+
|
|
285
|
+
---
|
|
286
|
+
|
|
287
|
+
### Critical Findings (action required before merge)
|
|
288
|
+
|
|
289
|
+
| # | Severity | Domain | File | Line(s) | Finding | Recommendation | Flagged By |
|
|
290
|
+
|---|----------|--------|------|---------|---------|----------------|------------|
|
|
291
|
+
|
|
292
|
+
### High-Priority Findings
|
|
293
|
+
|
|
294
|
+
| # | Severity | Domain | File | Line(s) | Finding | Recommendation | Flagged By |
|
|
295
|
+
|---|----------|--------|------|---------|---------|----------------|------------|
|
|
296
|
+
|
|
297
|
+
### Medium & Low Findings
|
|
298
|
+
|
|
299
|
+
| # | Severity | Domain | File | Line(s) | Finding | Recommendation |
|
|
300
|
+
|---|----------|--------|------|---------|---------|----------------|
|
|
301
|
+
|
|
302
|
+
### Praise (things done well)
|
|
303
|
+
<positive observations from reviewers>
|
|
304
|
+
|
|
305
|
+
---
|
|
306
|
+
|
|
307
|
+
### Cross-Review Notes
|
|
308
|
+
<any points of debate or disagreement between reviewers, with resolution>
|
|
309
|
+
|
|
310
|
+
---
|
|
311
|
+
|
|
312
|
+
### Reviewer Verdicts
|
|
313
|
+
|
|
314
|
+
| Reviewer | Verdict | Critical | High | Medium | Low | Info |
|
|
315
|
+
|----------|---------|----------|------|--------|-----|------|
|
|
316
|
+
| Security | APPROVE/REQUEST_CHANGES | N | N | N | N | N |
|
|
317
|
+
| Performance | APPROVE/REQUEST_CHANGES | N | N | N | N | N |
|
|
318
|
+
| Correctness | APPROVE/REQUEST_CHANGES | N | N | N | N | N |
|
|
319
|
+
|
|
320
|
+
### Overall Verdict: <APPROVE / REQUEST_CHANGES>
|
|
321
|
+
|
|
322
|
+
<one-paragraph summary: key risks, recommended actions, and overall assessment>
|
|
323
|
+
```
|
|
324
|
+
|
|
325
|
+
### Step 4: Cost Notice
|
|
326
|
+
|
|
327
|
+
Print a brief cost notice after the report:
|
|
328
|
+
|
|
329
|
+
```
|
|
330
|
+
Note: Team review used ~3x the tokens of a single-reviewer run (3 parallel reviewers + debate round).
|
|
331
|
+
```
|
|
332
|
+
|
|
333
|
+
---
|
|
334
|
+
|
|
335
|
+
## Rules
|
|
336
|
+
|
|
337
|
+
- This command is **read-only** — it MUST NOT create, modify, or delete any files
|
|
338
|
+
- All three reviewers run as Agent Teams teammates, not as Agent tool subagents
|
|
339
|
+
- If Agent Teams is unavailable at runtime (API error, version mismatch), fall back to running three sequential Agent tool subagents with `subagent_type: sr-reviewer` and skip the debate phase. Print a warning about the fallback.
|
|
340
|
+
- The debate phase is limited to one round per reviewer to control token costs
|
|
341
|
+
- Findings MUST include file paths and line numbers — vague findings are not acceptable
|
package/update.sh
CHANGED
|
@@ -94,7 +94,7 @@ if [[ -z "$CUSTOM_ROOT_DIR" && -f "$SCRIPT_DIR/install.sh" && -d "$SCRIPT_DIR/te
|
|
|
94
94
|
echo -e "${YELLOW}⚠${NC} You're running the updater from inside the specrails source repo."
|
|
95
95
|
echo -e " specrails updates a ${BOLD}target${NC} repository, not itself."
|
|
96
96
|
echo ""
|
|
97
|
-
read -p " Enter the path to the target repo (or 'q' to quit): " TARGET_PATH
|
|
97
|
+
read -p " Enter the path to the target repo (or 'q' to quit): " TARGET_PATH || TARGET_PATH="q"
|
|
98
98
|
if [[ "$TARGET_PATH" == "q" || -z "$TARGET_PATH" ]]; then
|
|
99
99
|
echo " Aborted. No changes made."
|
|
100
100
|
exit 0
|
|
@@ -107,7 +107,7 @@ if [[ -z "$CUSTOM_ROOT_DIR" && -f "$SCRIPT_DIR/install.sh" && -d "$SCRIPT_DIR/te
|
|
|
107
107
|
}
|
|
108
108
|
if [[ ! -d "$REPO_ROOT/.git" ]]; then
|
|
109
109
|
echo -e "${YELLOW}⚠${NC} Warning: $REPO_ROOT does not appear to be a git repository."
|
|
110
|
-
read -p " Continue anyway? (y/n): " CONTINUE_NOGIT
|
|
110
|
+
read -p " Continue anyway? (y/n): " CONTINUE_NOGIT || CONTINUE_NOGIT="n"
|
|
111
111
|
if [[ "$CONTINUE_NOGIT" != "y" && "$CONTINUE_NOGIT" != "Y" ]]; then
|
|
112
112
|
echo " Aborted. No changes made."
|
|
113
113
|
exit 0
|
|
@@ -658,7 +658,7 @@ except Exception:
|
|
|
658
658
|
fi
|
|
659
659
|
|
|
660
660
|
local answer
|
|
661
|
-
read -p " Regenerate agents? (y/N): " answer
|
|
661
|
+
read -p " Regenerate agents? (y/N): " answer || answer="n"
|
|
662
662
|
if [[ "$answer" == "y" ]] || [[ "$answer" == "Y" ]]; then
|
|
663
663
|
NEEDS_SETUP_UPDATE=true
|
|
664
664
|
ok "Will regenerate agents via /setup --update"
|