healthy-short-circuit-skip-llm
Bypass an expensive LLM diagnosis when an alert is provably healthy (resolved/info severity, no error annotations, evidence-investigated-and-empty), guarded…
Skill by kjuhwa
Chalk Labs
Need help deploying healthy-short-circuit-skip-llm?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/kjuhwa-healthy-short-circuit-skip-llm
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: ml, agent-building
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/healthy-short-circuit-skip-llm
What's inside
- Skill name
healthy-short-circuit-skip-llm- Description
- Bypass an expensive LLM diagnosis when an alert is provably healthy (resolved/info severity, no error annotations, evidence-investigated-and-empty), guarded behind an env-var kill switch in case the heuristic misfires.
- License
- MIT
- Instructions
- 52 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# Healthy Short-Circuit: Skip the LLM
## When to use
Many alerts are noise (scheduled health checks, "resolved" notifications). Calling a reasoning LLM on each one wastes tokens and adds latency. You can prove an alert is non-actionable using four cheap checks; only when all four pass do you skip the LLM and emit a deterministic "healthy" report.
## How it works
- All four conditions must hold:
1. `state ∈ {"normal","resolved","ok"}` (covers Grafana, CloudWatch, PagerDuty).
2. `severity ∈ {"info","none",""}` (rules out resolved-critical that still warrants RCA).
3. No error-signal annotations populated (`error`, `error_message`, `log_excerpt`, `failed_steps`).
4. At least one investigation evidence key is present in `evidence` (proves we actually queried the systems and found nothing).
- A `HEALTHY_SHORT_CIRCUIT` env var (default `"true"`) lets ops disable the optimization without a deploy if it ever misfires.
- The "evidence key present even if empty" check is the trick — empty `grafana_logs: []` is itself a health signal ("we queried Loki, no errors").
## Example
```python
_HEALTHY_STATES = frozenset({"normal","resolved","ok"})
_HEALTHY_SEVERITIES = frozenset({"info","none",""})
_ERROR_ANNOTATION_KEYS = ("error","error_message","log_excerpt","failed_steps")
_INVESTIGATED_EVIDENCE_KEYS = frozenset({
"grafana_logs", "grafana_metrics", "grafana_alert_rules",
"aws_cloudwatch_metrics", "aws_rds_events", "datadog_logs",
"datadog_monitors", "eks_pods", "eks_events", ...
})
def is_clearly_healthy(raw_alert, evidence) -> bool:
if not isinstance(raw_alert, dict): return False
state = str(raw_alert.get("state","")).lower().strip()
if state not in _HEALTHY_STATES: return False
labels = raw_alert.get("commonLabels", raw_alert.get("labels", {})) or {}
severity = str(labels.get("severity", raw_alert.get("severity",""))).lower().strip()
if severity not in _HEALTHY_SEVERITIES: return False
annotations = raw_alert.get("commonAnnotations", raw_alert.get("annotations", {})) or {}
if any(annotations.get(k) for k in _ERROR_ANNOTATION_KEYS): return False
return any(k in evidence for k in _INVESTIGATED_EVIDENCE_KEYS)
def diagnose_root_cause(state):
if _short_circuit_enabled() and is_clearly_healthy(state.get("raw_alert", {}),
state.get("evidence", {})):
return _handle_healthy_finding(state, ...) # no LLM call
# else proceed with normal LLM diagnosis
```
## Gotchas
- Always document the "blast radius if this misfires" — for a healthy short-circuit, the worst case is a real incident gets reported as healthy. Defend with the severity gate (firing critical never satisfies condition 2) plus the env kill switch.
- Use `frozenset` for the membership lookups — they're hot path.
- The "key present" check is subtly different from "key truthy". `grafana_logs: []` is a positive signal; `evidence.get("grafana_logs")` would treat it as missing.From kjuhwa/skills-hub, MIT.
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 17
- Last commit (this folder)
- Apr 25, 2026 (7 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/kjuhwa/skills-hub
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get healthy-short-circuit-skip-llm running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
healthy-short-circuit-skip-llm skill. Agent Mart. https://agentmart.dev/skills/kjuhwa-healthy-short-circuit-skip-llm (last updated 2026-10-24)