Borrowing it
Nothing to install: this file belongs to ForgeyClap/claude-forge. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/ForgeyClap/claude-forge/main/.claude/skills/forge-voice/SKILL.mdgit clone --depth 1 https://github.com/ForgeyClap/claude-forgeWrote 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/forgeyclap/claude-forge/forge-voice)<a href="https://agentmods.dev/skills/forgeyclap/claude-forge/forge-voice"><img src="https://agentmods.dev/badge/skills/forgeyclap/claude-forge/forge-voice.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 | $0.00056 | $0.01304 |
| Opus 5 | $0.00028 | $0.00652 |
| Sonnet 5 | $0.00011 | $0.00261 |
| Haiku 4.5 | $0.00006 | $0.00130 |
Grade A, and why
forge-voice 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 4d 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 — 43 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Forge playbook — AI voice / phone agent
A voice agent is a real-time, spoken, one-shot channel — the honesty + safety rules matter more than usual because a caller can't scroll back. RAG/knowledge answers defer to forge-rag (source-grounded, honest fallback); notifications/e-mail after a call defer to forge-integration (draft-gated, idempotent). Forge context: Dutch client phone agents (booking / reception / callback).
Hard rules
- Dutch-first when configured for a Dutch client. The caller-facing conversation is Dutch (natural, spoken Dutch — not translated-English cadence); female/Dutch voice when the owner asks. Code/comments stay English. Confirm language + voice per project.
- One question at a time. Ask a single question, wait, listen, then proceed — never stack multiple asks in one turn. Keep prompts short and spoken-natural; design the call as a small set of workflow states, each with one job to do.
- Natural turn-taking + barge-in. The caller can interrupt; the agent stops speaking within ~200ms of a real interruption. Keep voice-to-voice latency tight (aim well under ~1s). Handle endpointing / silence gracefully; short, polite closing phrases; no endless loop — end the call cleanly.
- AI disclosure + recording consent (EU/NL). Disclose clearly and early that the caller is speaking with an AI (EU AI Act Art. 50, in force Aug 2026). If the call is recorded, get a valid, specific, informed basis — a vague "this call may be recorded" is not enough under GDPR/AVG; record the legal basis. Honour opt-out immediately.
- One email per call maximum; idempotency before any notification. At most one outbound email per call, sent only after an idempotency check (a retry/duplicate turn must not fire a second mail). Consent-aware lead capture only.
- No outbound calls and no SMS without explicit owner approval. Inbound answering is the default safe mode; outbound dialing, SMS, and any customer traffic are owner-gated, irreversible actions. No autonomous dialing.
- Honest fallback — no hallucinated business facts. If the agent doesn't know (price, availability, policy), it says so and offers a human handoff / callback — never invents a business answer on a live call. Secrets (telephony/LLM keys) in env only.
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.
- 4d ago First seen · 43 lines · 56 tokens per session scan A 7d58f70e86ae
forge-voice is a skill published in the GitHub repository ForgeyClap/claude-forge (2 stars, last pushed 1mo ago), licensed MIT. It adds 56 tokens to every session and 1,304 once invoked, about $0.0003 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 2026-09-03.
Other skills, from other repositories
flame-harness-evaluator
Phase 6 — skeptical QA. Run the game, watch it, then judge against the contract. Default = functional check; --strict adds quality and edge-case passes.
flame-harness-admob
Phase 7 — analyze the game, decide a rewarded-ad strategy, guide manual AdMob ad-unit creation, and inject googlemobileads + ATT/UMP consent code.
flame-harness-screenshot
Phase 9 — capture store screenshots in the game's configured locales via integrationtest (ads hidden), fill ASO keywords, and upload via fastlane.
flame-harness-submit
Phase 10 — upload store text metadata + categories via fastlane, then pause with exact manual steps for the final iOS review submission and Android production promotion.
flame-harness-contract
Phase 4 — propose verifiable completion criteria and mandatory hard gates; reach AGREED (1-pass default, multi-round negotiation in --strict).
flame-harness-design
Phase 3 — define the Flutter designtokens.dart spec (palette, typography, spacing), the game's art/visual concept, and the asset/audio sourcing plan.