debug

An eight-step debugging process for Untether, a project that operates software across a fleet of systems. It supports either a broad sweep of logs and open issues or a detailed investigation of named GitHub issues.

In plain words
What is it for?
Use it to triage fleet-wide problems, investigate specific Untether issues, consult project and upstream records, identify systemic patterns, and prepare verification details.
Why use it?
It organizes evidence gathering and hypothesis testing so investigations do not stop at guesses or symptoms. It can also define how a proposed fix should be verified.

Command for Claude Code

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 commands/littlebearapps/untether/debug
Clone the repo
git clone --depth 1 https://github.com/littlebearapps/untether

Made for: Claude Code.

Per session 37 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 4,166 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. 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 $0.00037 $0.04166
Opus 5 $0.00018 $0.02083
Sonnet 5 $0.00007 $0.00833
Haiku 4.5 $0.00004 $0.00417

Measured yesterday against content hash 678d419cb02d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

debug scanned grade A with 1 finding 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.

Runs shell commandslowCapability

Expected in a hook, worth knowing in a rule or an instructions file.

(e.g. `event=subprocess.liveness_stall`) and time window.
.claude/commands/debug.md · 327 lines

How it starts

The opening of the file, as written. The whole thing — 327 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are handling the user's /debug request. /debug is Untether's 8-step deep debugging protocol. It runs in one of two modes — general sweep (ranked triage report across fleet logs + open issues, no auto-filing) or targeted (full 8-step rigor per named GitHub issue, posted as an issue comment with a verification spec).

User input: $ARGUMENTS

This command is project-scoped to Untether. It is not the generic /debug <issue> shipped in the user's global ~/.claude/commands/debug.md; project commands take precedence when invoked from inside /home/nathan/untether.

Companion files (canonical references)

File Step covered Purpose
.claude/commands/debug/README.md n/a How the bundle fits together
.claude/commands/debug/step-classify.md Step 1 15 Untether issue classes + diagnostic hints
.claude/commands/debug/step-evidence.md Step 2 Data-source catalogue (journalctl, structlog, fleet SSH, MCP, state files)
.claude/commands/debug/step-research.md Step 3 Docs, closed issues, upstream engine repos, library docs
.claude/commands/debug/systemic-patterns.md Step 4 15-20 known Untether patterns + memory-aware exceptions
.claude/commands/debug/step-fix.md Step 7 Implementation checklist (tests, lint, format, changelog, branch model)
.claude/commands/debug/step-verify.md Step 8 Post-fix verification (dev restart, integration tests, attestation, fleet rollout)
.claude/commands/debug/output-template.md output Debug-report-comment and triage-report templates
~/.claude/commands/monitor/severity-rubric.md (reused) Severity buckets and label routing — single source of truth
~/.claude/commands/monitor/signal-categories.md (reused) Bug/enhancement signal taxonomy

Read the relevant companion file BEFORE executing each step. Do not duplicate their content here — they are the canonical source.

Sub-commands

Form Action
/debug (no args) General sweep, default window = 24 hours
/debug sweep [hours] General sweep, custom window
/debug <issue_number> [more numbers...] Targeted: full 8-step per issue
/debug --issue <n> Targeted: single issue (explicit form)
/debug --help Print usage block

Read the full file on GitHub · 327 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 · 327 lines · 37 tokens per session scan A 678d419cb02d

Subscribe to this mod's changes

debug is a command published in the GitHub repository littlebearapps/untether (66 stars, last pushed 2d ago), licensed MIT. It adds 37 tokens to every session and 4,166 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 1 finding (runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.