Agent Stack Audit

Agent Stack Audit

Bottom line: Run this monthly. Your agent stack accumulates dead weight faster than you think — zombie crons, unused APIs still billing you, scripts referencing killed strategies, and duplicate tools doing the same job. This skill finds all of it in one pass.

When to Invoke

Scheduled: First Monday of every month, 05:00 local time. Quick scan: Every Sunday (cron health only — 5 min). Manual trigger: Any time the stack feels bloated, bills are unexpected, or reliability has degraded. Trigger phrases: "stack audit", "clean up my automations", "what's still running?", "am I paying for anything unused?", "why are my API costs high?"

Audit Scope (6 Categories)

1. Crons — Are they alive and earning their keep?

For each launchd agent / cron job / scheduled task:

• Is the process actually running? (check PID, plist/cron status)
• When did it last fire successfully?
• What does it produce? Is that output being consumed by anything?
• Is there a newer/better approach that renders this obsolete?

Questions to answer:

• "This cron runs every hour. Has its output file been read in the last 30 days?"
• "This watchdog monitors a bot that was killed 2 months ago — is the watchdog still running?"

2. Scripts — Dead code?

Scan your automation directories for Python/shell scripts:

• Last modified date vs last executed date
• Scripts referencing killed bots or cancelled APIs
• Scripts built for old projects that are now closed
• Duplicate scripts doing the same job

3. API Keys — Are you paying for something unused?

Cross-reference your API inventory against actual script usage:

• Any API key configured but never called in the last 30 days?
• Any paid subscription that maps to zero active script usage?
• Any free-tier key that's been maxed out — is an upgrade worthwhile?

Common culprits: data providers, news APIs, notification services, AI APIs at old models.

4. Skills — Superseded or never used?

Review your installed skills:

• Any skill built but never actually invoked?
• Any skill replaced by a newer, better version?
• Any skill with overlapping functionality that could be merged?

5. Memory Files — Stale project context?

Review project memory and context files:

• Any project memory not updated in 30+ days?
• Projects marked "on hold" for 60+ days with no activity?
• Contradictions between your main context file and individual project files?

6. Tools & Subscriptions — Better option available?

Using web search:

• "Is [tool] still the best option for [use case] as of [current month]?"
• Check for: price drops on existing tools, new free tiers, open source alternatives, better APIs
• Flag if a paid tool now has a free/cheaper replacement

Classification Framework

For each item found, classify as:

Output Format

File: state/stack_audit_YYYY-MM-DD.md

# Stack Audit — [DATE]
_Scope: Crons, Scripts, APIs, Skills, Memory, Tools_
_Items scanned: [N] | Issues found: [M] | Recommended actions: [K]_

---

## 🔴 RECOMMEND DELETE (zombies)
| Item | Type | Reason | Risk if deleted |
|---|---|---|---|
| old_watchdog.sh | Cron | Bot killed Mar 12, watchdog still running | None — bot is dead |

## 🟡 RECOMMEND REPLACE/MERGE (redundant)
| Item | Type | Replaced By | Action |
|---|---|---|---|
| old_intelligence_scan.py | Script | Tech Scout skill | Kill script, activate skill |

## 🟠 RECOMMEND UPGRADE (outdated)
| Item | Type | Current | Better Option | Est. Saving/Gain |
|---|---|---|---|---|
| image_gen API call | API | DALL-E 3 | [Cheaper/better alternative] | ~40% cost reduction |

## 🟢 HEALTHY — No Action
[List with one-line confirmation each]

## ⚪ NEEDS USER INPUT
[Items that require your decision — presented as yes/no questions]

---

## Summary
- Recommended deletes: [N items] — saves [X $/mo or compute]
- Recommended upgrades: [N items]
- Estimated cleanup time if approved: ~[X hours]
- Most important action: [single highest-leverage item]

Execution Rules

1. Never delete anything without explicit user approval — only RECOMMEND
2. One exception: If a script references a dead API that has been confirmed cancelled → safe to comment out the call and log it. Don't delete the file.
3. If unsure → classify as ⚪ UNKNOWN. Never guess on deletions.
4. After approvals: execute cleanup and write state/stack_cleanup_YYYY-MM-DD.md confirming what was removed and what the before/after count was.

Quick Scan (Sunday Cron Health Check)

Lighter weekly version — runs in 5 minutes:

1. Check all scheduled tasks/crons are firing on schedule
2. Check all PID files match running processes
3. Check for any ERROR or FAILED in last 24h logs
4. Output: state/cron_health_YYYY-MM-DD.md — just a pass/fail list, no detail

Why This Matters

Left unchecked, automation stacks accumulate:

• Zombie crons that run every hour and log to files no one reads
• API subscriptions billing monthly for services killed 6 months ago
• Duplicate scripts that were "temporary" but became permanent
• Stale context files that feed wrong information to your agents

The monthly audit is what separates a clean, reliable stack from a pile of debt that fails when you need it most. Most teams discover 20-40% of their crons are either dead or redundant on the first pass.