ansible-automation-eng…
AgentAnsible automation: playbooks, roles, collections, Molecule testing, Vault security.
**Scope**: Detecting, stopping, and recovering from agent death loops.
Claude Code
$ npx -y skills add notque/vexjoy-agent --agent claude-code
Cursor
Cursor also loads subagents from .claude/agents, so copy this file there to use it in Cursor.
> **Scope**: Detecting, stopping, and recovering from agent death loops.
---
| Situation | Signal | Action | |-----------|--------|--------| | Same error 3+ consecutive | Identical error message/stack | STOP, BLOCKERS.md, root cause analysis | | Agent modifying same lines | STATUS.md shows identical targets | Trigger intervention, change strategy | | Compilation + lint cycling | Build fails after lint; lint breaks after build | Enforce compilation-first | | Context at 70%+ | Spawning cost growing | PROGRESS.md, compress context |
---
AGENT: golang-general-engineer TASK: Fix channel direction linting errors ATTEMPT 1 — Result: FAILED — 3 channel direction errors in consumer.go ATTEMPT 2 — Changed directions → FAILED — compilation broken in producer.go ATTEMPT 3 — Reverted, changed producer.go → FAILED — original errors returned ATTEMPTS: 3/3 FAILED — LOOP DETECTED PATTERN: Fix in consumer breaks producer; fix in producer reverts consumer ACTION: Manual intervention — design issue, not lint fix
---
1. ASSIGN: code-modification-agent — SUCCESS: `go build ./...` exits 0 2. VERIFY: `go build ./...` — must pass before step 3 3. ASSIGN: lint task — SUCCESS: lint exits 0 AND `go build ./...` still exits 0
---
## Blocker: Channel Direction Death Loop AGENT: golang-general-engineer — 3/3 FAILED ROOT CAUSE: consumer.go and producer.go share a channel in types.go. types.go must change first — neither file fixable independently. REQUIRED STRATEGY: 1. Modify types.go channel type 2. Verify compilation 3. Fix consumer.go and producer.go together ESCALATION: User approval needed to modify types.go (shared interface)
---
**Detection**: `grep -A2 "ATTEMPT" STATUS.md | grep "Error:" | sort | uniq -d` **Fix**: Before retry, identify what changed. If answer is "nothing", stop and document in BLOCKERS.md.
**Detection**: `grep -B5 "lint\|ruff\|golangci" HANDOFF.md | grep -i "success criteria"` **Fix**: Compilation verification as first AND last step of every lint task:
SUCCESS CRITERIA: 1. Compilation passes 2. Lint passes 3. Compilation still passes after lint
**Detection**: `grep -c "RETRY\|Re-assign\|Try again" HANDOFF.md` **Fix**: Re-dispatch requires: (1) specific error from prior attempt, (2) root cause identified, (3) new strategy with file/line scope.
---
| Pattern | Fix | |---------|-----| | Same error 3 consecutive entries | STOP, root cause analysis, change approach | | Build passes → lint → build fails | Enforce compilation-first: verify build after lint | | 3/3 without BLOCKERS.md | Create BLOCKERS.md entry before further coordination | | Agent assigned file in another domain | Check STATUS.md assignments before dispatching | | Context summarization mid-task | Write PROGRESS.md at 70% threshold |
---
# Repeated identical errors grep -A2 "ATTEMPT" STATUS.md | grep "Error:" | sort | uniq -d # Retry directives without strategy (should be 0) grep -c "RETRY\|Re-assign\|Try again" HANDOFF.md # Lint before compilation in task sequence grep -B5 "lint\|ruff\|golangci" HANDOFF.md | grep -i "success criteria" # Completed loops grep -rn "ATTEMPTS: 3/3" . --include="STATUS.md"
Essays and writing behind this toolkit live at vexjoy.com. VexJoy Agent connects plain-English requests to specialist agents, skills, and workflows. /do selects the knowledge and tools needed for your task.
Ansible automation: playbooks, roles, collections, Molecule testing, Vault security.
**Scope**: Module selection patterns, builtin vs command/shell decisions, collection modules, and version-specific module changes…
**Scope**: Molecule test scenarios, ansible-lint rules, idempotency validation, and check-mode patterns **Version range**:…
Universal rules injected by /do at dispatch. Each agent's .md file supplies domain rules.
**Scope**: Failure modes in agent output style — over-reporting, self-congratulation, verbose narration, and hedging. Covers what…
Zero-dependency combat visual upgrades: CSS particle replacement, Framer Motion combat juice, CSS 3D card transforms.
Surgical 1-2 file edit. Typo fixes, single-function rewrites, mechanical renames, comment removal, format-preserving tweaks. Hard…