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.
npx agentmods add instructions/soleimanmansouri/debug-bank/claude-mdgit clone --depth 1 https://github.com/soleimanmansouri/debug-bankWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/soleimanmansouri/debug-bank/claude-md)<a href="https://agentmods.dev/instructions/soleimanmansouri/debug-bank/claude-md"><img src="https://agentmods.dev/badge/instructions/soleimanmansouri/debug-bank/claude-md.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00799 | $0.00799 |
| Opus 5 | $0.00400 | $0.00400 |
| Sonnet 5 | $0.00160 | $0.00160 |
| Haiku 4.5 | $0.00080 | $0.00080 |
Grade A, and why
debug-bank CLAUDE.md 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 5d ago.
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.
How it starts
The opening of the file, as written. The whole thing — 77 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Debug Bank — Debug Trajectory Protocol
Debugging Protocol
Follow this exact chain for every debugging session. Each step produces evidence for the next — never skip.
Before Investigating
- Pattern Check — Read
patterns/for matching root cause type. If a match exists, verify the known fix applies before re-investigating. This takes 30 seconds and saves hours. - Domain Search — Search your domain catalogs (if any) for similar symptoms.
The Trajectory
- Reproduce — Get the exact error. Capture full output (logs, stack trace, HTTP status). If you can't reproduce, add logging and try again.
- Hypothesize — State 2-3 possible root causes ranked by likelihood. Each hypothesis must be falsifiable with a specific test.
- Isolate — Test hypotheses one at a time. Use binary search: disable half the system, check if bug persists, narrow.
- Diagnose — Identify the single root cause. Trace the full call chain from trigger to symptom. Check for "root cause behind the root cause."
- Fix — Minimal change. Verify the fix addresses the root cause, not a symptom.
- Record — Add the trajectory to your domain catalog using this format:
### [Category] Short Title (date) - **Symptom:** What the user/system saw - **Root cause:** The actual technical cause - **Fix:** What was changed (file:line) - **Key insight:** The generalizable lesson - **Pattern:** P-number if it matches an existing pattern
The 3-Exchange Rule
If 3 rounds of iterative fixing show no progress: STOP. Do not continue the same approach.
Instead:
- Re-plan from scratch with everything you now know
- Add logging/instrumentation to gather new evidence
- Switch strategy entirely (e.g., from code-level fix to config-level fix)
Anti-Patterns (Enforced)
- Never fix a symptom without finding the root cause
- Never apply a fix you saw work elsewhere without verifying it matches the current root cause
- Never keep iterating past 3 exchanges — switch strategy
- Never assume you understand the bug from the description alone — always reproduce first
- Always check if 2+ symptoms share one root cause before filing separate fixes
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.
- 5d ago First seen · 77 lines · 799 tokens per session scan A 7e3528c944dc
debug-bank CLAUDE.md is an instructions file published in the GitHub repository soleimanmansouri/debug-bank (1 stars, last pushed 3mo ago), licensed MIT. It adds 799 tokens to every session, about $0.0040 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.
Other instructions, from other repositories
meta-llm-charter CLAUDE.md
Instructions for entropyvortex/meta-llm-charter, covering meta v3.1 core charter, bias, meta-0, r1 decompose and r2 ask gate.
awesome-reviewers CLAUDE.md
Instructions for baz-scm/awesome-reviewers, covering awesome reviewers — repository guidelines, what this project is, source of truth, site layer and machine interface.
code-context AGENTS.md
AGENTS.md instructions for infino-ai/code-context, covering code-context: notes for ai agents, project overview, repo map, build, test, gates and conventions.
gtm-cheat-codes AGENTS.md
Instructions for zapier/gtm-cheat-codes, covering agents.md and project overview.
Stata-CLI AGENTS.md
Instructions for Utolaris/Stata-CLI, covering agents.md, project structure and working rules.
specgate CLAUDE.md
Claude Code instructions for thanhtung2693/specgate: This file configures Claude Code when it contributes to the SpecGate repository. It is not product guidance for people using SpecGate.