vrr/spawn
2
0
Fork 0
mirror of https://github.com/OpenRouterTeam/spawn.git synced 2026-04-28 03:49:31 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 10

feat(agents): add style-reviewer teammate and auto-update Claude Code (#2685)

Browse source 
Add a new style-reviewer agent to the refactor team that enforces project
rules from CLAUDE.md and .claude/rules/ (biome lint, shell script compat,
type safety, test conventions). Runs proactively during refactor cycles.

Also add `claude update --yes` to all 4 launcher scripts (refactor.sh,
discovery.sh, security.sh, qa.sh) so agents always run on the latest
Claude Code version before each cycle.

Co-authored-by: Claude <claude@anthropic.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
A 2026-03-16 00:41:17 -07:00 • committed by GitHub
parent c9c662aaba
commit ab51b09a03
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 42 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

4
.claude/skills/setup-agent-team/discovery.sh
View file

@ -95,6 +95,10 @@ if [[ ! -f "${MANIFEST}" ]]; then
exit 1
fi
# Update Claude Code to latest version before launching
log_info "Updating Claude Code..."
claude update --yes 2>&1 | tee -a "${LOG_FILE}" || log_warn "Claude Code update failed (continuing with current version)"
export CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1
# Persist into .spawnrc so all Claude sessions on this VM inherit the flag
if [[ -f "${HOME}/.spawnrc" ]]; then

4
.claude/skills/setup-agent-team/qa.sh
View file

@ -261,6 +261,10 @@ if [[ "${RUN_MODE}" == "soak" ]]; then
fi
fi
# Update Claude Code to latest version before launching
log "Updating Claude Code..."
claude update --yes 2>&1 | tee -a "${LOG_FILE}" || log "WARNING: Claude Code update failed (continuing with current version)"
# Launch Claude Code with mode-specific prompt
# Enable agent teams (required for team-based workflows)
export CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1

28
.claude/skills/setup-agent-team/refactor-team-prompt.md
View file

@ -183,7 +183,31 @@ Assign teammates to labeled issues first (no plan mode). Remaining teammates do
**NEVER close a PR** — only the security team can close PRs. If a PR is stale, broken, or superseded, comment explaining the issue and move on.
**NEVER touch human-created PRs** — only interact with PRs that have `-- refactor/` in their description.
6. **community-coordinator** (Sonnet)
7. **style-reviewer** (Sonnet) — Best match for `style` or `lint` labeled issues. Proactive: enforce project rules and conventions from `CLAUDE.md` and `.claude/rules/`.
**Proactive scan procedure:**
1. Run `bunx @biomejs/biome check src/` in the CLI package — fix all violations (lint, format, grit rules like `no-type-assertion`)
2. Check shell scripts against `.claude/rules/shell-scripts.md`:
- No `echo -e` (must use `printf`)
- No `source <(cmd)` in curl-piped scripts
- No `((var++))` with `set -e`
- No `set -u` (use `${VAR:-}` instead)
- No `python3 -c` / `python -c` (use `bun -e` or `jq`)
- No relative paths for sourcing (`source ./lib/...`)
3. Check TypeScript against `.claude/rules/type-safety.md`:
- No `as` type assertions (except `as const`)
- No `require()` / `module.exports` (ESM only)
- No manual multi-level typeguards (use valibot)
- No `vitest` imports (use `bun:test`)
4. Check test files against `.claude/rules/testing.md`:
- No `homedir` from `node:os` (use `process.env.HOME`)
- No subprocess spawning (`execSync`, `spawnSync`, `Bun.spawn`)
- Tests must import from real source, not copy-paste functions
Only create a PR if actual violations are found. ONE PR max, fixing all violations found.
Run `bunx @biomejs/biome check src/` and `bun test` after every change to confirm fixes don't break anything.
7. **community-coordinator** (Sonnet)
First: `gh issue list --repo OpenRouterTeam/spawn --state open --json number,title,body,labels,createdAt`
**COMPLETELY IGNORE issues labeled `discovery-team`, `cloud-proposal`, or `agent-proposal`** — those are managed by the discovery team. Do NOT comment on them, do NOT change labels, do NOT interact in any way. Filter them out:
@ -220,7 +244,7 @@ Assign teammates to labeled issues first (no plan mode). Remaining teammates do
## Commit Markers
Every commit: `Agent: <agent-name>` trailer + `Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>`
Values: security-auditor, ux-engineer, complexity-hunter, test-engineer, code-health, pr-maintainer, community-coordinator, team-lead.
Values: security-auditor, ux-engineer, complexity-hunter, test-engineer, code-health, pr-maintainer, style-reviewer, community-coordinator, team-lead.
## Git Worktrees (MANDATORY)

4
.claude/skills/setup-agent-team/refactor.sh
View file

@ -151,6 +151,10 @@ if [[ "${RUN_MODE}" == "refactor" ]]; then
log "Pre-cycle cleanup done."
fi
# Update Claude Code to latest version before launching
log "Updating Claude Code..."
claude update --yes 2>&1 | tee -a "${LOG_FILE}" || log "WARNING: Claude Code update failed (continuing with current version)"
# Launch Claude Code with mode-specific prompt
# Enable agent teams (required for team-based workflows)
export CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1

4
.claude/skills/setup-agent-team/security.sh
View file

@ -206,6 +206,10 @@ done
log "Pre-cycle cleanup done."
# Update Claude Code to latest version before launching
log "Updating Claude Code..."
claude update --yes 2>&1 | tee -a "${LOG_FILE}" || log "WARNING: Claude Code update failed (continuing with current version)"
# Launch Claude Code with mode-specific prompt
# Enable agent teams (required for team-based workflows)
export CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1