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/clawcall-dev/clawcall/skillnpx skills add ClawCall-Dev/ClawCall --skill skillgit clone --depth 1 https://github.com/ClawCall-Dev/ClawCallWhat 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 | $0.00096 | $0.03725 |
| Opus 5 | $0.00048 | $0.01862 |
| Sonnet 5 | $0.00019 | $0.00745 |
| Haiku 4.5 | $0.00010 | $0.00373 |
Grade A, and why
clawcall 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 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.
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 — 312 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ClawCall
ClawCall lets you make real US phone calls for the user. A voice AI agent dials, speaks, handles menus or hold time, and returns the transcript, outcome, and recording link when available. The first outbound call can auto-provision an API key.
Base URL: https://api.clawcall.dev
Core Rule
The phone agent only knows the Call instructions you send as task. More relevant detail is better. Build a complete briefing before calling, and do not make the user supply public/business facts you can reasonably look up yourself.
Choose The Workflow
| User intent | Do this |
|---|---|
| Call someone now | Build rich Call instructions, POST /call, then poll GET /call/{call_id} until lifecycle = "finalized". |
| Get through to a person / connect me | Use outbound calling with bridge_number and a handoff trigger in the Call instructions. |
| Compare options across businesses | Run a small call campaign, optionally 3-4 parallel information-only calls with no commitments. |
| Set up my ClawCall profile/personality | Configure global voice/personality/greeting and, if needed, inbound answering profile. |
| Configure how my number answers calls | Use inbound profile setup. Do not POST /call. |
| What calls came in? | Poll inbound history with GET /me/calls?direction=inbound.... |
| Link this agent to my ClawCall account | Use the saved API key to produce the sign-in link. |
| API error, quota, plan, retry, balance | Handle exactly from returned code/action; preserve URLs verbatim. |
Product Coaching
Educate at decision points, not as a generic pitch.
- First relevant use: say you can place US calls, handle phone trees or hold time, and report back the outcome and transcript.
- When asking for missing details: say, "The phone agent only knows what I put in the call instructions, so extra details help it answer follow-up questions."
- Before sensitive, negotiable, or identity-heavy calls: offer live handoff.
- Before complex calls, surface likely verification, OTP, payment, fee, or live-decision points and offer the right call plan.
- For option searches, offer to call several places and compare without committing unless the user gave clear approval boundaries.
- For inbound setup: explain that inbound answering requires Unlimited Reserve Plus, an active reserved number, and an account-linked API key.
- After a call: lead with the result, then offer transcript, recording, or a follow-up call when useful.
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.
- evals/pressure-scenarios.md 5.1 KB
- references/account-linking-and-data.md 3.1 KB
- references/api-contract.md 5.5 KB
- references/errors-and-limits.md 3.9 KB
- references/examples.md 8.7 KB
- references/inbound-reserved-numbers.md 5.5 KB
- references/outbound-calls.md 9.0 KB
- references/profile-and-personality.md 3.9 KB
- SkillCard.md 5.1 KB
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.
- yesterday First seen · 312 lines · 96 tokens per session scan A 35731eb50bc6
clawcall is a skill published in the GitHub repository ClawCall-Dev/ClawCall (7 stars, last pushed 7d ago), licensed MIT. It adds 96 tokens to every session and 3,725 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 2026-08-31.
Other skills, from other repositories
agentbox-info
Spin up isolated sandboxes ("boxes") for coding agents, run them in parallel, queue background runs with -i, and push commits safely through the host relay. Use when the user wants to run Claude Code / Codex / OpenCode in a sandbox, start more boxes, attach to a running box, or otherwise operate the agentbox CLI on…
agentbox-setup
Generate an agentbox.yaml for the current AgentBox workspace. Invoke when the user opens a sandbox without an agentbox.yaml or asks to (re)configure one.
agentbox
Fork the current agent session into a new VM or local Docker container with all the project files, agent settings and session teleported into.
release
Prepare, validate, tag, publish, and monitor guarded PwrSnap desktop releases. Use when the user asks to release PwrSnap, select alpha/beta/maintenance-candidate/RC/stable versions, prepare a vX.Y.Z or vX.Y.Z-prerelease tag, update release notes or CHANGELOG.md for a desktop release, verify package.json/tag/changelog…
e2e-docker-repro
Run and debug PwrSnap Desktop E2E tests inside the local Docker/xvfb harness that mirrors GitHub Actions. Use when investigating GHA Desktop E2E failures, Linux/xvfb-only Electron bugs, flaky Playwright specs, teardown hangs, source-filter flakes, or when asked to reproduce a PR CI failure locally with…
release-download-stats
Check and summarize PwrSnap GitHub Release asset download statistics. Use when the user asks for download counts, bytes served, DMG stats, mac updater ZIP traffic, Windows installer traffic, per-release stats such as beta.20/beta.18, or whether GitHub release downloads show any traffic.