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 skills/elumenotion/guideants/audiocppnpx skills add Elumenotion/GuideAnts --skill audiocppgit clone --depth 1 https://github.com/Elumenotion/GuideAntsWrote 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/skills/elumenotion/guideants/audiocpp)<a href="https://agentmods.dev/skills/elumenotion/guideants/audiocpp"><img src="https://agentmods.dev/badge/skills/elumenotion/guideants/audiocpp.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 | $0.00107 | $0.01499 |
| Opus 5 | $0.00053 | $0.00749 |
| Sonnet 5 | $0.00021 | $0.00300 |
| Haiku 4.5 | $0.00011 | $0.00150 |
Grade A, and why
audiocpp-extended 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.
This is a copy
100% identical to audiocpp-extended — 204 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 140 lines — stays where its author put it; the contents beside it link to each section on GitHub.
audio.cpp extended (experimental)
Paths — fixed layout, do not probe or re-derive. The sandbox CWD is the
notebook's output directory. Skill scripts live under Skills/audiocpp/scripts/
relative to it. Write deliverables with bare filenames; never prefix with Output/.
GuideAnts product ASR/TTS only expose {text, voice, speed} against a fixed
catalog. This skill reaches the raw engine surface without GuideAntsApi /
ServiceModes changes.
Default path: PC sandbox → GPU host raw audiocpp gateway. That host is a
transparent reverse proxy to full audiocpp_server (/asr/*, /tts/*,
/private/*), plus /files staging and /admin/* for model fetch / private
spawn. Scripts use AUDIOCPP_SKILL_BASE_URL when set. Do not call
127.0.0.1:18082/18084 from a PC sandbox — those ports exist only inside the
GPU host AI container.
Everything here is experimental. Run the probe first, trust its report, and tell the user plainly when a route is blocked.
Environment (required for PC → the GPU host)
AUDIOCPP_SKILL_BASE_URL=http://<gpu-host-lan-ip>:8112/audiocpp-skill
AUDIOCPP_SKILL_TOKEN=<same as the GPU host GA_AUDIOCPP_SKILL_TOKEN>
Optional: HF_TOKEN for gated downloads (runs on the GPU host).
With these set, scripts stage workspace audio via /files, call raw engine
JSON under /asr|/tts|/private, download models into /models-local/skill/…
on the GPU host, and spawn private engines there. See references/engine-api.md.
Always start with the probe
python3 Skills/audiocpp-extended/scripts/probe.py
Expect routes.route5_remote_skill_gateway.open: true on a PC sandbox. Do not
attempt a route the probe marked blocked.
Voice cloning consent rule
Cloning from a reference clip is supported when the speaker consents (own voice or stated permission). Decline only unconsented third-party imitation.
What to run (raw gateway path)
ASR / TTS must already be loaded on the GPU host via GuideAnts Settings (API lifecycle). Then:
What ships with it
9 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- Output/gateway_test.wav 379 KB
- references/deferred-models.md 8.2 KB
- references/engine-api.md 3.5 KB
- scripts/diarize.py 28 KB runs code
- scripts/engine_tool.py 8.2 KB runs code
- scripts/fetch_model.py 7.0 KB runs code
- scripts/probe.py 11 KB runs code
- scripts/skill_gateway_client.py 6.8 KB runs code
- scripts/spawn_engine.py 11 KB runs code
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 Changed · -46 lines · -33 tokens per session scan D → A 7e0b385ec958
- 6d ago First seen · 186 lines · 140 tokens per session scan D 4cdbce99b021
audiocpp-extended is a skill published in the GitHub repository Elumenotion/GuideAnts (56 stars, last pushed 2d ago), licensed Apache-2.0. It adds 107 tokens to every session and 1,499 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. It is 100% identical to audiocpp-extended, differing in 204 lines, and is treated as a copy.
Other skills, from other repositories
project-release
Prepare, publish, verify, or recover a cloakbrowser-mcp release only when the user explicitly requests release work. Require a Prompt MCP-confirmed target version and stability, follow the repository's version, changelog, PR, GitHub Release, npm, Docker, MCP Registry, and docs process, and never tag, publish…
project-pull-request
Create, update, prepare, or review a cloakbrowser-mcp GitHub Pull Request only when the user explicitly requests PR work. Follow the repository's branch, base/head, assignee, title, template, check, security, and reporting rules; review remains read-only, and commit, push, PR creation/update, readiness changes, merge…
spec-driven-development
Author the authoritative /spec workflow for a substantial cloakbrowser-mcp change when the user invokes /spec, explicitly requests a specification, or authorizes specification work that lacks a durable contract. Establish repository facts, use the globally installed Prompt MCP for every material decision, persist…
planning-and-task-breakdown
Author the authoritative /plan workflow for substantial cloakbrowser-mcp work when the user invokes /plan or explicitly requests implementation planning, decomposition, milestones, or review-gated task packets. Require an approved specification, use Prompt MCP for unresolved material planning and authorization…
code-review-and-quality
Review a proposed or completed cloakbrowser-mcp change only when the user explicitly requests code review, quality review, or a pre-merge assessment. Evaluate strict TypeScript, CLI and transport contracts, upstream-tool parity, child-process and session boundaries, tests, docs, packaging, workflows, security, and…
documentation-and-adrs
Write or revise cloakbrowser-mcp technical documentation or record a settled architecture decision only when the user explicitly requests documentation, an ADR, or documentation required by an authorized public-contract change. Use for bridge architecture, CLI and environment configuration, stdio or Streamable HTTP…