fable-debug

A step-by-step method for investigating bugs, unexplained errors, failing tests, or incorrect output.

In plain words
What is it for?
Debugging software failures, narrowing down causes, checking whether a fix really worked, and avoiding repeated failed attempts.
Why use it?
It replaces guess-and-check fixes with reproduction, explicit hypotheses, targeted tests, and evidence about what changed.

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/debabsah/fable-method/fable-debug
Any agent
npx skills add debabsah/fable-method --skill fable-debug
Clone the repo
git clone --depth 1 https://github.com/debabsah/fable-method

Made for: Claude Code, Codex.

Per session 85 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 767 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 $0.00085 $0.00767
Opus 5 $0.00043 $0.00383
Sonnet 5 $0.00017 $0.00153
Haiku 4.5 $0.00009 $0.00077

Measured yesterday against content hash aee2a1380292, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

fable-debug 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 yesterday.

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.

skills/fable-debug/SKILL.md · 29 lines

How it starts

The opening of the file, as written. The whole thing — 29 lines — stays where its author put it; the contents beside it link to each section on GitHub.

fable-debug

A bug is a wrong belief, not just a wrong line. Fix your model of the system first; the code change then falls out. Never debug by mutation — "change something, rerun, hope" burns evidence and moves the bug around.

The procedure

  1. Reproduce first. Get the minimal command that shows it red, and run it. A failure you can't trigger on demand can't be verified fixed — and if you can't reproduce it, that is the finding: report it and instrument (logging, a capture) instead of shipping a speculative patch.
  2. State the contradiction in one sentence. "X should produce Y because Z; it produces W." If you can't fill in Z, you don't understand the intended behavior yet — go read the source of truth first (R1).
  3. Run a hypothesis ledger, cheapest-first. List 2–5 candidate causes. Each gets a designed probe with a predicted outcome written down before running it — a probe whose outcome you can't predict is a coin flip, not an experiment. One variable per probe. Strike out falsified rows in writing; that's what prevents circling back to a cause you already killed.
  4. Bisect the delta when a working/broken pair exists — version, commit, input, environment, config. Differential evidence beats staring at code.
  5. Three strikes → up a level. The same category of fix failing twice means the shared assumption underneath is wrong (R2). Stop patching; re-read the contradiction; widen the frame.
  6. Fix the invariant, not the instance. Root cause found → grep every caller and sibling path; the fix goes where all paths route through, or the siblings stay broken.
  7. Verify red → green on the exact reproduction from step 1 — red before the fix, green after, at the layer of the claim (→ fable-verify). A fix you never watched fail isn't proven to be the fix.
  8. Root-cause the escape. Always the second question: why did the existing checks miss this? Mint the runnable rule — the test/assert/check that would have caught it — and log the gotcha (trap → cause → rule) to .fable/project.md. If a task file is open in .fable/tasks/, record the decision trail there. And if the broken behavior was vouched for earlier, find that claim in .fable/claims-logand in .fable/claims-log.<year>, where fable-ship archives older ones; a claim you can't find is not a claim that was never made — then mark it FALSIFIED <date> — <what actually broke>: a falsified Verified: is a calibration miss, and how the check missed it is the gotcha.

Read the full file on GitHub · 29 lines

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. yesterday First seen · 29 lines · 85 tokens per session scan A aee2a1380292

Subscribe to this mod's changes

fable-debug is a skill published in the GitHub repository debabsah/fable-method (1 stars, last pushed 1mo ago), licensed MIT. It adds 85 tokens to every session and 767 once invoked, about $0.0004 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 2026-08-31.

Related

Other skills, from other repositories

issue-triage

Issue triage: audit open issues, categorize, detect duplicates, cross-ref PRs, risk assessment, post comments. Args: "all" for deep analysis of all, issue numbers to focus (e.g. "42 57"), "en"/"fr" for language, no arg = audit only in French.

rtk-ai/rtk · 70 tokens

hud

Configure HUD display options (layout, presets, display elements).

Yeachan-Heo/oh-my-claudecode · 14 tokens

moai-ref-ui-polish

UI polish and interface-completion reference: the small visual details — concentric border radius, optical alignment, shadow-vs-border, motion easing, typography smoothing, tabular numbers, icon stroke weight, hit areas — that separate polished interfaces from generic ones. Agent-extending skill that amplifies…

modu-ai/moai-adk · 98 tokens

kelos

Author, debug, and operate Kelos resources (Task, Session, Workspace, AgentConfig, TaskSpawner) on Kubernetes. Use for Kelos CRDs, resource manifests, the kelos CLI, or live cluster operations. Do not use for ordinary Kelos repo code edits, tests, reviews, build/CI, or git tasks unless they involve those resources.

kelos-dev/kelos · 76 tokens

trellis-brainstorm

Guides collaborative requirements discovery before implementation. Creates task directory, seeds PRD, asks high-value questions one at a time, researches technical choices, and converges on MVP scope. Use when requirements are unclear, there are multiple valid approaches, or the user describes a new feature or complex…

mindfold-ai/Trellis · 65 tokens

mulmoterminal-keys

Bind keyboard shortcuts and fix keyboard/clipboard behaviour in MulmoTerminal. Writes keymap, which Settings cannot set at all — its Keyboard shortcuts section is read-only, listing every action bound or not plus a send row. Explains copyOnSelect, questionPaneEnabled and terminalSubmit, which have their own Settings…

receptron/mulmoterminal · 289 tokens