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 commands/littlebearapps/untether/qagit clone --depth 1 https://github.com/littlebearapps/untetherWhat 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 | $0.00105 | $0.02852 |
| Opus 5 | $0.00053 | $0.01426 |
| Sonnet 5 | $0.00021 | $0.00570 |
| Haiku 4.5 | $0.00011 | $0.00285 |
Grade A, and why
qa 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 2d 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 — 186 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are handling /qa. /qa is Untether's validation stage — it answers one
question: "is this target validated enough for its risk?" It classifies the
target, picks the lightest sufficient depth, runs the checks, and — only when
asked — drives the integration-test tiers against the allowlisted dev bot. It
is the VALIDATE boundary between PREPARE (/plan→/implement or /fix) and
RELEASE (/pr-dev / /pr-main).
User input: $ARGUMENTS
Untether adaptations (read first)
Load .claude/rules/workflow-commands.md (routing + the 7 cross-cutting rules).
docs/reference/integration-testing.md is the design-of-record — /qa
implements its tier definitions, chat IDs, and MCP tool list; it never
re-litigates them. Key points for /qa:
- Authority: observe + exercise, never change. Read repo/logs/diffs, run
safe local tests, drive the bounded, allowlisted dev bot, file capped
findings, write the attestation marker. Never fix code, merge, tag,
release, roll the fleet, or mutate a protected file. Repair routes to
/debug→/fix. - Dev/staging separation. Drive
@untether_dev_bot(untether-dev.service) only. Never@hetz_lba1_bot(staging), never a fleet/production target. Neversystemctl restartfrom inside this session. - Confirm-gated + idempotent. Findings are confirm-gated (surface the drafted
gh issue create, de-dupe open and closed first); a re-invoked/qamust not double-file or double-write the marker. - Redaction. Scrub tokens/keys/env/chat-content/fleet identifiers from any
log/response evidence before it lands in a finding (see
.claude/commands/debug/step-evidence.md). - Untether-mode.
AskUserQuestion/ExitPlanModereturn empty — the--runconfirmation is stated in text and STOPS for a reply. Keep the report brief.
The live-bot guardrails (non-negotiable)
A slash command cannot enforce capability limits in code, so these are hard rules the command obeys:
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.
- 2d ago First seen · 186 lines · 105 tokens per session scan A bd2dd5a1668a
qa is a command published in the GitHub repository littlebearapps/untether (66 stars, last pushed 3d ago), licensed MIT. It adds 105 tokens to every session and 2,852 once invoked, about $0.0005 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 2026-08-30.
Other commands, from other repositories
refresh-compare-pages
Freshly research, add, and refresh marketing compare pages, then open a descriptive PR.
release
This compatibility command keeps /oh-my-claudecode:release available without loading the full release skill description in every Claude Code session.
research
Research a technical or product question.
fd-verify
Verify the implementation — run tests, check regression on affect.md files, review and scan; blocks /fd-done on failure.
wtfp:review-section
Review a drafted section for evidence, citations, coherence, requirements, and reader expectations.
hello-stranger
Say hello stranger.