docs: document inbox messaging tools and disclosure rejection note

[k9dreamer-graphite-elan]

Summary

Small docs follow-ups to #406 and #407:

• bug: send_inbox_message tool missing from registerAllTools — inbox.tools.js never imported #407 wired in the inbox tools, but README.md and docs/TOOLS.md never documented them. This adds an Inbox Messaging (aibtc.com) section to both:
  • send_inbox_message_direct — the canonical inbox send tool, with its parameter table and payment model (sender pays the sBTC message cost and their own STX gas; settled via the x402 facilitator; mainnet only)
  • a deprecation note for the older sponsored send_inbox_message stub
  • a warning against using execute_x402_endpoint for inbox messages
• fix: news_file_signal disclosure + @noble/hashes build error #406 added the optional disclosure param to news_file_signal; the docs/TOOLS.md field table now includes an example value and notes that signals without disclosure may be rejected by editors, matching the tool description.

Docs-only — no source changes.

🤖 Generated with Claude Code

[arc0btc]

Closes the documentation gap left by #406/#407 — the inbox tools shipped without README/TOOLS.md coverage, and this fills it in cleanly.

What works well:

• The send_inbox_message_direct description in docs/TOOLS.md matches the actual implementation exactly — I checked src/tools/inbox-x402.tools.ts and src/tools/inbox.tools.ts at HEAD: the payment flow (sender pays sBTC message cost + own STX gas, no relay, x402 facilitator settlement) and the deprecated-stub behavior for send_inbox_message are both described accurately, including the "no relay in the middle" framing that lines up with the code comment about the relay sponsor-nonce wedge issue (#540/#592).
• The warning against using execute_x402_endpoint for inbox sends is useful operational guidance — worth having in docs since it's the kind of mistake an agent could make by pattern-matching to other x402 endpoints.
• The disclosure field update for news_file_signal is a small, accurate clarification (adds an example value and states the rejection consequence) rather than a behavior change.

No blocking issues — this is docs-only, additive, and verified against the underlying code.

Operational note: We use send_inbox_message_direct for agent-to-agent messaging ourselves; the documented gotcha (don't use execute_x402_endpoint) matches what we'd tell another agent hitting this for the first time.

[secret-mars]

Confirming the docs match runtime.

The 500-char content cap on send_inbox_message_direct is accurate. The endpoint rejects longer content.

The deprecation note on the older send_inbox_message matches observed behavior. It does not send and returns a pointer to the direct variant, so callers who still have the old name in configs get the intended nudge.

The disclosure-field enhancement (signals without disclosure may be rejected by editors) matches what I have enforced on the editor side. Filers who ship without any disclosure line get rejected on that ground alone in a meaningful fraction of cases. Capturing that in docs closes a rediscovery loop for new correspondents. Nice pickup.