Skip to content
Agentmart
Submit a tool

debug

A check fails or behavior is wrong and the cause isn't obvious — reproduce, isolate the root cause, fix narrowly, and prove it.

Skill by bryan824

IndexedMIT
Write a review

Chalk Labs

Need help deploying debug?

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.

Book a call

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/bryan824-debug

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.

We use your email only for this newsletter. Privacy

About

A check fails or behavior is wrong and the cause isn't obvious — reproduce, isolate the root cause, fix narrowly, and prove it.

Use cases: productivity

No 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/debug

What's inside

Skill name
debug
Description
A check fails or behavior is wrong and the cause isn't obvious — reproduce, isolate the root cause, fix narrowly, and prove it.
License
MIT
Instructions
35 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Debug

Find the causal defect, not just a patch that hides the symptom.

- Capture a pass/fail signal for the reported behavior. Preserve the original
  scenario while shrinking a reproduction; the small case must still explain it.
  For intermittent failures, record conditions and observed frequency instead of
  imposing an arbitrary flake threshold. If reproduction is unavailable, seek
  decisive traces and name what remains unproven.
- Trace callers, state, guards, sinks and configuration. Compare known-good
  behavior and recent changes when available. Rank plausible explanations by
  what observable evidence would distinguish them, then test the most informative
  one. Change one causal variable at a time where possible.
- Keep probes identifiable. Probes and their cleanup need permission to write,
  even for files you created. An approved tool call does not widen which files
  you may change. Prefer
  existing tests, in-memory checks or focused traces over new files and noisy
  logging; ask before creating an extra file. Measure performance against a
  comparable baseline. Redact secrets before sharing commands, logs or artifacts,
  and say when redaction limits diagnosis.
- Fix the shared cause where it belongs and inspect sibling callers. Do not
  compensate separately in every path or keep retrying an unchanged theory.
  Repeated failure calls for a new hypothesis, decomposition or escalation—not a
  success verdict because an attempt budget ran out.
- Re-run the exact original signal and relevant regression checks. Assert the
  visible result, not merely that nothing threw. Inspect the real rendered/output
  surface for visual or generated-artifact bugs; synthetic checks alone leave that
  requirement unverified. If no honest test seam exists, report the design gap.

Deliver: cause, distinguishing evidence, fix and regression results, removed
probes and remaining uncertainty. Hand the uncommitted change to independent
verification. Keep a durable lesson in its existing owner only when it changes
current guidance; no mandatory debug diary.

From bryan824/kirin-pi, MIT.

More from bryan824/kirin-pi

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Oct 3, 2026 (29 days ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from bryan824

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 debug 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.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying debug?

CallGet help
Cite this page

debug skill. Agent Mart. https://agentmart.dev/skills/bryan824-debug (last updated 2026-10-24)