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/r00tify/crashdx/claude-mdgit clone --depth 1 https://github.com/r00tify/crashdxWrote 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/r00tify/crashdx/claude-md)<a href="https://agentmods.dev/instructions/r00tify/crashdx/claude-md"><img src="https://agentmods.dev/badge/instructions/r00tify/crashdx/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 | $0.01347 | $0.01347 |
| Opus 5 | $0.00674 | $0.00674 |
| Sonnet 5 | $0.00269 | $0.00269 |
| Haiku 4.5 | $0.00135 | $0.00135 |
Grade A, and why
crashdx 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 — 98 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
Commands
swift build # debug build (crashdx, crashdx-mcp, CrashDXCore)
swift build -c release
swift test # full test suite (Swift Testing, not XCTest)
swift test --filter DiagnosisEngineTests # one suite (regex match on the suite name)
swift test --filter watchdogTimeoutFiresOnSyntheticFixture # one @Test function
swift run crashdx analyze <report.ips> [--json --tier full] [--dsym <path>] [--no-spotlight] [--no-archives]
swift run crashdx symbolicate <report.ips> [--dsym <path>] [--no-spotlight] [--no-archives]
swift run crashdx-mcp # stdio MCP server
python3 Tests/mcp-smoke.py [.build/debug/crashdx-mcp] # manual MCP protocol smoke test;
# NOT part of `swift test`; run after `swift build`
Architecture
Three targets share one core:
CrashDXCore, the library:.ipsparsing (IPSFile), dSYM discovery (DSYMLocator), symbolication (Symbolicator), the diagnosis engine (Diagnosis/), and untrusted-text rendering (SafeRendering). Every report-derived string a front end prints must go throughsanitized(_:). Imports only Foundation.crashdx: CLI.analyze/symbolicatesubcommands.crashdx-mcp: MCP server (stdio, newline-delimited JSON-RPC). The only target importing an external package (modelcontextprotocol/swift-sdk). Note SwiftPM resolves per package, so library consumers still pull that graph.
AnalyzePipeline (Sources/CrashDXCore/AnalyzePipeline.swift) is the single shared
parse → locate dSYMs → symbolicate → diagnose → build-report sequence. Both the CLI and
the MCP server call it. Never duplicate this flow in either front end; add new steps
there so the two can't drift.
Diagnosis engine
Three deterministic stages, no LLM calls (full spec in docs/DESIGN.md):
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 · 98 lines · 1,347 tokens per session scan A ded2a4491b70
crashdx CLAUDE.md is an instructions file published in the GitHub repository r00tify/crashdx (7 stars, last pushed 1mo ago), licensed MIT. It adds 1,347 tokens to every session, about $0.0067 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
DebugMCP AGENTS.md
AGENTS.md instructions for microsoft/DebugMCP, covering agent guidelines for debugmcp, project overview, architecture, key components and documentation maintenance.
sentry AGENTS.md
AGENTS.md instructions for getsentry/sentry, covering sentry development guide for ai agents, command execution guide, python command execution requirements, backend development commands and (not needed for pytest). highly recommended.
airbroke AGENTS.md
AGENTS.md instructions for icoretech/airbroke, covering project knowledge base, structure, where to look, code map and conventions.
keynobi AGENTS.md
AGENTS.md instructions for thiagodmont/keynobi, covering agent instructions, tech stack, before you write code, key rules and testing instructions.
django-orbit CLAUDE.md
Instructions for astro-stack/django-orbit, covering claude.md, what this is, commands, install in editable mode with dev dependencies and run all tests.
sentry CLAUDE.md
Claude Code instructions for getsentry/sentry, a project described as: Developer-first error tracking and performance monitoring.