Borrowing it
Nothing to install: this file belongs to SpekoAI/mcp-dev-calls. 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/SpekoAI/mcp-dev-calls/main/AGENTS.mdgit clone --depth 1 https://github.com/SpekoAI/mcp-dev-callsWrote 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/instructions/spekoai/mcp-dev-calls/agents-md)<a href="https://agentmods.dev/instructions/spekoai/mcp-dev-calls/agents-md"><img src="https://agentmods.dev/badge/instructions/spekoai/mcp-dev-calls/agents-md/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/instructions/spekoai/mcp-dev-calls/agents-md"><img src="https://agentmods.dev/badge/instructions/spekoai/mcp-dev-calls/agents-md.svg" alt="Reviewed on agentmods" width="80" 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 | $0.02868 | $0.02868 |
| Opus 5.5 | $0.01147 | $0.01147 |
| Sonnet 5.5 | $0.00574 | $0.00574 |
| Haiku 4.5 | $0.00287 | $0.00287 |
Grade A, and why
mcp-dev-calls AGENTS.md 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 9d 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 — 198 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — @spekoai/mcp-calls (speko)
Agent-oriented guide for using this package correctly. It places real, disclosed phone calls to real people and businesses, and it does terminal speech (TTS/STT). Calls cost money. Read the Safety section before you dial. Running in a headless cloud agent sandbox? See docs/agent-platforms.md.
1. What speko is
One binary, two personalities, decided by how it's launched (see mcp/src/cli/router.ts):
- MCP server — when an MCP host (Claude Code, Claude Desktop, any client) spawns it over stdio (piped, non-TTY stdin, no subcommand). It exposes 6 tools.
- CLI — when you run it in a terminal with a subcommand (
speko audio ...,speko voices,speko dnc ...,speko usage,speko call ..., etc.).
stdout/stderr invariant (do not break it): in MCP-server mode stdout is reserved
for JSON-RPC; all logs, progress, and diagnostics go to stderr. Every CLI
subcommand runs its handler and process.exit()s before the MCP server is ever
constructed, which is what keeps stdout clean. A typo'd subcommand prints usage to
stderr and exits 2 — it never silently boots a server that would hang the caller.
2. Two ways to use it
(a) As an MCP server — the 6 tools
Prefer these tools over shelling out when you're an agent inside an MCP host.
| Tool | Kind | Use when |
|---|---|---|
lookup_business(name, location?, phone_number?, utc_offset_minutes?) |
read-only | You want the verified-directory path for a business. Resolves to dialable candidates and mints a short-lived signed dial_token per callable one. This is the only path that can authorize make_call. If you already found the official number via web search, pass phone_number (E.164) to skip the directory lookup — it's still carrier-verified as a business line. |
make_call(dial_token, objective, caller_name, context?, behavior?, greet_first?, after_hours_confirmation?, max_duration_seconds?, language?, wait?) |
mutating | Place the disclosed, objective-scoped call authorized by a dial_token. Blocks until the call finishes, returns the OUTCOME + transcript with honest connected/answered/not_connected. wait:false returns a call ID to poll with get_call. |
call_number(phone_number, objective, caller_name, recipient_name?, context?, behavior?, greet_first?, utc_offset_minutes?, after_hours_confirmation?, max_duration_seconds?, language?, wait?) |
mutating | The default path once you have a number — business or personal, mobiles allowed. Works with just the user's Speko key, no directory/carrier keys. Only dial a number the user gave you or you found — never one you invented. wait:false returns a call ID to poll with get_call. |
call_me(message, mode?, context?, after_hours_confirmation?, max_duration_seconds?, wait?) |
mutating | Ring this install's locally verified owner. There is no destination field. Use notify for one-way delivery and converse for a reply. A converse instruction is usable only when the result says `confirmation: confirmed |
check_call_readiness() |
read-only | Preflight before the first call, or when calling doesn't work. Reports auth, prepaid credit balance, outbound caller-ID readiness, owner verification (call_me.available), and the enforced client profile, each with a concrete next step. Never dials. |
get_call(call_id) |
read-only | Re-check a call's status, connected/answered, OUTCOME, and transcript. Use after a timeout, or to inspect a finished call. Never dials. |
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.
- 9d ago Changed · +2 lines · +147 tokens per session 53d6ccf17647
- 2mo ago First seen · 196 lines · 2,721 tokens per session scan A fb57bb2de763
mcp-dev-calls AGENTS.md is an instructions file published in the GitHub repository SpekoAI/mcp-dev-calls (4 stars, last pushed 5d ago), licensed MIT. It adds 2,868 tokens to every session, about $0.0115 per session on Opus 5.5. 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.
mcp-dev-calls AGENTS.md compared with instructions from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| mcp-dev-calls AGENTS.mdSpekoAI/mcp-dev-calls | Instructions file | 4 | 5d ago | MIT | A | 2,868 | Codex, OpenCode |
| vscode buildNext.instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 9,719 | GitHub Copilot |
| next.js AGENTS.mdvercel/next.js | Instructions file | 143k | yesterday | MIT | A | 7,461 | Codex, OpenCode |
| langchain AGENTS.mdlangchain-ai/langchain | Instructions file | 149k | yesterday | MIT | A | 4,680 | Codex, OpenCode |
| vscode oss-third-party-notices.instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 5,516 | GitHub Copilot |
| vscode copilot-instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 3,093 | GitHub Copilot |
| deepseek-harness AGENTS.mddeepseek-ai/deepseek-harness | Instructions file | 274k | 2d ago | MIT | A | 4,313 | Codex, OpenCode |