deflake

A test-analysis helper that examines failed GitHub Actions runs, the automated checks that run in GitHub, to find tests that fail intermittently.

In plain words
What is it for?
Use it to rank flaky tests by failure frequency, review their failed runs, and plan ways to make the worst offenders reliable.
Why use it?
It helps separate flaky tests from real bugs and infrastructure problems, so teams can focus on the tests causing the most repeated failures.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/stacklok/toolhive/deflake
Any agent
npx skills add stacklok/toolhive --skill deflake
Clone the repo
git clone --depth 1 https://github.com/stacklok/toolhive

Made for: Claude Code, Codex.

Per session 48 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,699 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00048 $0.01699
Opus 5 $0.00024 $0.00849
Sonnet 5 $0.00010 $0.00340
Haiku 4.5 $0.00005 $0.00170

Measured 2d ago against content hash a0967721a1c0, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

deflake scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

The scan reads SKILL.md. This mod also ships 1 executable file (collect-flakes.py), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/skills/deflake/SKILL.md · 158 lines

How it starts

The opening of the file, as written. The whole thing — 158 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Deflake Tests

Discovers, ranks, and plans fixes for flaky tests by analyzing GitHub Actions failures on main.

Arguments

/deflake                    # Full analysis: discover, rank, and plan fixes
/deflake --report           # Report only: show flake rankings without planning fixes
/deflake --top N            # Analyze and plan fixes for the top N flakes (default: 3)

Phase 1: Collect and Rank Flakes

Run the collection script. It handles all deterministic data collection and aggregation. If CI log formats change over time, update the script directly.

python3 .claude/skills/deflake/collect-flakes.py

The script outputs three sections:

  1. FLAKE REPORT — overall stats (total runs, failure rate, date range)
  2. RANKED FAILURES — table sorted by failure count with job, mode, and test name
  3. FAILURE DETAILS — per-test breakdown with links to each failed run

Phase 1 complete

Read the script output and use it directly for the report. The LLM's only job in this phase is to categorize each entry as a flake, real bug, or infra issue:

  • Flake: Appears multiple times intermittently, interspersed with successful runs
  • Real bug: Appeared after a specific commit and every run after that failed until a fix landed. Check git log for related fixes
  • Infra flake: Entries tagged [INFRA] by the script, or failures with mode connection refused / infra

Phase 2: Present the Report

Present the script output as a formatted report. Add categorization (flake / real bug / infra) to each entry. Example format:

## Flake Report — main branch

**Period**: 2026-04-01 to 2026-04-10
**Runs analyzed**: 23 total, 8 failed (35% failure rate)

### Top Flaky Tests

| Rank | Test | Job | Failures | Failure Mode |
|------|------|-----|----------|--------------|
| 1 | Workload lifecycle ... [It] should track ... | E2E (api-workloads) | 5/23 | timeout (120s) |
| 2 | ... | ... | ... | ... |

### Real Bugs (not flakes)
- [Test name] — Introduced by [commit], fixed by [commit/PR]

### Infra Failures
- [N] runs failed due to [description]

Read the full file on GitHub · 158 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 158 lines · 48 tokens per session scan A a0967721a1c0

Subscribe to this mod's changes

deflake is a skill published in the GitHub repository stacklok/toolhive (2,056 stars, last pushed yesterday), licensed Apache-2.0. It adds 48 tokens to every session and 1,699 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

network-rca

Kubernetes network root cause analysis skill powered by Kubeshark MCP. Use this skill whenever the user wants to investigate past incidents, perform retrospective traffic analysis, take or manage traffic snapshots, extract PCAPs, dissect L7 API calls from historical captures, compare traffic patterns over time, detect…

kubeshark/kubeshark · 163 tokens

kfl

KFL2 (Kubeshark Filter Language) reference. This skill MUST be loaded before writing, constructing, or suggesting any KFL filter expression. KFL is statically typed — incorrect field names or syntax will fail silently or error. Do not guess at KFL syntax without this skill loaded. Trigger on any mention of KFL, CEL…

kubeshark/kubeshark · 203 tokens

security-audit

Kubernetes network security audit skill powered by Kubeshark MCP. Use this skill whenever the user wants to audit a cluster for security threats, detect compromised workloads, find malicious traffic patterns, hunt for indicators of compromise (IOCs), check for data exfiltration, identify C2 (command and control)…

kubeshark/kubeshark · 182 tokens

install

Kubeshark installation and deployment skill. Use this skill whenever the user wants to install Kubeshark, deploy Kubeshark to a Kubernetes cluster, set up Kubeshark, configure Kubeshark helm values, generate a Kubeshark config file, customize Kubeshark deployment, troubleshoot Kubeshark installation, upgrade…

kubeshark/kubeshark · 257 tokens

qa-find-bugs

Drive the live product at https://dashboard.bex.co as a signed-in QA user (QAEMAIL / QAPASSWORD from .env), hunt real bugs across the hosting features, research each fix down to file:line, file the non-duplicate findings to the w6 board through /pm, and /ship the scheduled milestone. Use when the user asks to QA the…

bex-co/bex · 96 tokens

pm

Inspect and maintain the repository's .pm workstreams, inbox notes, milestones, and tasks. Use when the user asks for PM board status or requests a supported board mutation such as creating, promoting, adding, or completing work.

bex-co/bex · 48 tokens