New: 97,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Debug

by guardana in guardana/guardana

Systematic diagnosis of a red test, a rule that fires or stays silent wrongly, a scan or probe whose artifact looks wrong, a collector error, a red CI run or a gate that is green for the wrong reason.

Apache-2.0Auto-check passedSecurity

Install Debug

skills CLI
$ npx skills add guardana/guardana --skill debug -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install guardana/guardana debug --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/guardana/guardana.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/debug .claude/skills/debug && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
debug
GitHub stars
152
Token cost
~933 tokens
SKILL.md length
515 words
Files
1
Skills in repo
16
Repo updated
First seen
Licence
Apache-2.0

At a glance

Systematic diagnosis of a red test, a rule that fires or stays silent wrongly, a scan or probe whose artifact looks wrong, a collector error, a red CI run or a gate that is green for the wrong reason.

  • Works in 4 steps: Reproduce or locate the evidence first.… → One hypothesis at a time, each with the… → Fix the cause where every path passes,… → …
  • Tasks that involve Prompt injection and agent security
  • SKILL.md covers Method, Where the evidence lives and Shapes that mislead here
  • Calls uv, gh and docker

What it does

Debug is an agent skill from guardana/guardana. Systematic diagnosis of a red test, a rule that fires or stays silent wrongly, a scan or probe whose artifact looks wrong, a collector error, a red CI run or a gate that is green for the wrong reason. Use before proposing any fix — it names where this project's evidence lives and the failure shapes that mislead here.

Its SKILL.md is about 930 tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Security, covering Prompt injection and agent security. It works with Docker and Python. The repository describes itself as: Open-source AI security verification for model artifacts, live endpoints, MCP servers, and recorded agent traces. Reproducible evidence for release decisions. The licence is Apache-2.0.

When your agent uses it

  • Tasks that involve Prompt injection and agent security

Example prompts

  • “/debug”

Requirements

  • Python 3
  • Docker

Workflow steps

4 steps, taken from the first numbered list in SKILL.md.

  1. Reproduce or locate the evidence first. No fix is proposed until the failing command, the
  2. One hypothesis at a time, each with the observation that would refute it. Check it with
  3. Fix the cause where every path passes, then add the test that would have caught it —
  4. Hand log reading and bulk lookups to scout / runner; keep the reasoning here.

What it can do on your machine

Read from SKILL.md and the folder at commit dd3920e. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • uv
    • gh
    • docker
    • python3

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use uv, gh and docker, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Debug loads about 933 tokens when it runs. Until then it costs about 81 tokens; SKILL.md has 515 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~81
When it runs · the whole SKILL.md, loaded when a task matches
~933

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from guardana/guardana at commit dd3920e, republished under its Apache-2.0 licence (© guardana). 515 words, ~933 tokens.

Download SKILL.mdSave it as .claude/skills/debug/SKILL.md (or your agent's skills folder).
name
debug
description
Systematic diagnosis of a red test, a rule that fires or stays silent wrongly, a scan or probe whose artifact looks wrong, a collector error, a red CI run or a gate that is green for the wrong reason. Use before proposing any fix — it names where this project's evidence lives and the failure shapes that mislead here.
argument-hint
[symptom, test id, rule id or run file]

Debug — evidence before theory

Symptom: $ARGUMENTS

Method

  1. Reproduce or locate the evidence first. No fix is proposed until the failing command, the exact error text and the input (fixture, run file, target) are known.
  2. One hypothesis at a time, each with the observation that would refute it. Check it with the cheapest read (one test with -x -vv, one JSON field, one log line) before reading code broadly.
  3. Fix the cause where every path passes, then add the test that would have caught it — including the negative case. A symptom patched in one command comes back through the others that share the code.
  4. Hand log reading and bulk lookups to scout / runner; keep the reasoning here.

Where the evidence lives

questionlook here
why is this test reduv run pytest <path>::<test> -x -vv; delete __pycache__ first after a same-size edit
does the rule fire / stay quiet / say inconclusiveuv run guardana rule test <rule-id>; the fixtures beside the rule; guardana.core.testing doubles
what did the command really writerun the documented command against a fixture or a fake endpoint (three lines of http.server) and open the JSON it wrote — the method that found the worst defect in three releases
what does a saved run containuv run guardana run inspect <file>; schemas/ for the contract
why did CI go redgh run view <id> --log-failed; read the failing step, not the job's conclusion
the collectordocker compose -f deploy/docker-compose.dev.yml up -d, then uv run guardana-collector migrate / serve; the PostgreSQL tests need GUARDANA_TEST_DATABASE_URL
an imageuv run --no-project python scripts/image_smoke.py (builds and runs both; needs docker)
the siteuv run python scripts/build_site.py --check; python3 -m http.server -d site 8099
Show full SKILL.md (233 more words)Show less

Shapes that mislead here

  • Green with skips. Without PostgreSQL ~245 tests skip and pytest exits 0; CI refuses the skip. Read the M skipped count, not the exit code.
  • A stale cache. .ruff_cache has answered for a changed file; __pycache__ reuses old bytecode after a same-size edit in the same second; a cached wheel in the isolated example runs hides the data files an extension change touches. scripts/ci_local.sh clears all three.
  • A green suite is not a working command. Every unit test can pass while the command drops a field, sends twice the budget or reports "no regression" over an indeterminate run — each of these has happened. Run the command; read the artifact.
  • Silence spelled pass. A check that could not run and returned clean looks like a passing check in every gate. The verdict must be inconclusive or a finding.
  • A test measuring an echo. An assertion on a log line, a document or a mock's call count measures what the code said. Move it to the seam where the value has to arrive.
  • A count in prose. A number that disagrees with the registry is the prose being stale, not the registry; generate_docs.py --check and sync_site.py --check say which.

Write down: symptom → evidence → cause → fix → the test that now guards it. If the lesson is general, add one line to the matching .claude/rules/ file; the story goes in the commit message.

© guardana, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .claude/skills/debug of guardana/guardana.

Open the folder on GitHubat commit dd3920e

Compare with similar skills

Debug next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Debug compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Debug this skillguardana/guardana152—~933Automated safety check: PassApache-2.0
OpenartAI45Lab/OpenART231—~918Automated safety check: NotesAGPL-3.0
Skylos Securityduriantaco/skylos843—~545Automated safety check: PassApache-2.0
Code Review Securitynicepkg/auto-company1921 repos~3.9kAutomated safety check: PassMIT
Cyber NeoHainrixz/cyber-neo281—~5.9kAutomated safety check: WarnMIT
Format Hak5 Pineapple Recon Data JSONsneakerhax/Arsenal112—~259Automated safety check: PassGPL-3.0

Similar skills

  • Openart

    AI45Lab/OpenART

    Guide an OpenART agent or contributor through planning, running, extending, and debugging the framework.

    231 GitHub stars~918 tokensUpdated 4 days ago
    SecurityAuto-check: notes

  • Skylos Security

    duriantaco/skylos

    Investigate and harden Skylos security behavior. An agent skill from duriantaco/skylos.

    843 GitHub stars~545 tokensUpdated yesterday
    SecurityAuto-check passed

  • Code Review Security

    nicepkg/auto-company

    Security-focused code review checklist and automated scanning patterns.

    192 GitHub starsUsed in 1 repo~3.9k tokens
    SecurityAuto-check passed

  • Cyber Neo

    Hainrixz/cyber-neo

    Comprehensive cybersecurity analysis for any local project. An agent skill from Hainrixz/cyber-neo.

    281 GitHub stars~5.9k tokensUpdated 2 mo ago
    SecurityAuto-check: warnings

  • Format Hak5 Pineapple Recon Data JSON

    sneakerhax/Arsenal

    Format and pretty-print Hak5 WiFi Pineapple recon JSON scan files for readability.

    112 GitHub stars~259 tokensUpdated 1 mo ago
    SecurityAuto-check passed

  • Format Hak5 Pineapple Recon Data Table

    sneakerhax/Arsenal

    Display Hak5 WiFi Pineapple recon scan data as a formatted table.

    112 GitHub stars~298 tokensUpdated 1 mo ago
    SecurityAuto-check passed

More from guardana/guardana

All 16 skills in this repo

  • Content Model

    guardana/guardana

    Route wording work to GPT (codex CLI) or Gemini (agy CLI) instead of writing it with Claude — landing-page copy, a readability rewrite of a README or docs page, attack and judge prompts for a rule…

    152 GitHub stars~1.2k tokensUpdated yesterday
    Auto-check passed

  • Ship

    guardana/guardana

    Commit and push a finished change the way this repo requires — explicit paths, one commit per logical change, no attribution, the five documentation places answered, the site checks before a push (a…

    152 GitHub stars~836 tokensUpdated yesterday
    Auto-check: notes

  • Add A Rule

    guardana/guardana

    Add security coverage to Guardana the way this repository requires — as a rule, evaluator or target, never by patching the engine — with the fixtures, the framework mapping and the documentation…

    152 GitHub stars~1.1k tokensUpdated yesterday
    Auto-check passed

  • Auto

    guardana/guardana

    Autonomous, non-interactive run of the whole lifecycle for one development task — size, plan, build, gate, review, fix, commit — without stopping for questions.

    152 GitHub stars~792 tokensUpdated yesterday
    Auto-check passed

  • Docs

    guardana/guardana

    Documentation work in this repo — the five places a user-visible change must answer, the page conventions the site build enforces, the tests that pin prose to the registry, and a simplification pass…

    152 GitHub stars~1k tokensUpdated yesterday
    Auto-check passed

  • False Green Audit

    guardana/guardana

    Hunt for the failure this project exists to prevent — code that compiles, types, tests green, and quietly reports "all clear" about something it never examined.

    152 GitHub stars~1.1k tokensUpdated yesterday
    Auto-check passed

Works with

Categories

Questions about Debug

What does Debug do?

Systematic diagnosis of a red test, a rule that fires or stays silent wrongly, a scan or probe whose artifact looks wrong, a collector error, a red CI run or a gate that is green for the wrong reason. Debug is an agent skill from guardana/guardana. Systematic diagnosis of a red test, a rule that fires or stays silent wrongly, a scan or probe whose artifact looks wrong, a collector error, a red CI run or a gate that is green for the wrong reason.

When should I use Debug?

Debug fits situations like: tasks that involve Prompt injection and agent security.

How do I install Debug in Claude Code?

Run `npx skills add guardana/guardana --skill debug -a claude-code`. Or copy the skill folder (.claude/skills/debug in guardana/guardana) into .claude/skills/debug in your project. Claude Code loads it when a task matches its description.

How do I install Debug in Codex?

Run `npx skills add guardana/guardana --skill debug -a codex`. Or copy the skill folder (.claude/skills/debug in guardana/guardana) into .agents/skills/debug in your project. Codex loads it when a task matches its description.

Can I use Debug in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add guardana/guardana --skill debug -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/debug, .gemini/skills/debug, .github/skills/debug and .opencode/skills/debug in your project.

What does Debug need to run?

Going by SKILL.md and its folder, Debug needs the command-line tools its instructions call (uv, gh, docker and python3). Our summary lists: Python 3; Docker.

Does Debug access the network?

SKILL.md contains no URLs. Its commands use uv, gh and docker, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Debug safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Debug use?

Debug is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Debug use?

About 933 tokens (SKILL.md is roughly 3.7k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Debug?

Skills that share tags, products or a category with Debug: Openart (AI45Lab/OpenART, 231 stars), Skylos Security (duriantaco/skylos, 843 stars), Code Review Security (nicepkg/auto-company, 192 stars) and Cyber Neo (Hainrixz/cyber-neo, 281 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Debug?

guardana (a GitHub organization) maintains it in guardana/guardana, which has 152 GitHub stars. The repository holds 16 skills in this directory. The repository was last updated on October 6, 2026.

Source: guardana/guardana on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.