OPP: Messaging platform permission modeling (Slack, Discord, Telegram)

Extend the policy prover to model messaging platform permissions for the most common agent communication channels: Slack, Discord, and Telegram.

Today OPP models binary capabilities and GitHub credential scopes. Real deployments connect agents to messaging platforms where the security question is not just "can traffic reach this host" but "what can the bot actually do" -- post messages vs upload files vs read DMs vs manage channels.

Scope

Add API descriptor and credential registry files following the existing pattern in crates/openshell-prover/registry/:

• Slack: Model OAuth bot scopes (chat:write, channels:read, files:write, etc.) to allowed API actions. Distinguish read-only scopes from write scopes. Model WebSocket endpoints as event-receive-only.
• Discord: Model bot permissions bitmask to allowed actions. Distinguish message send from file upload, channel management, DM access.
• Telegram: Model Bot API methods to read vs write actions. Distinguish sendMessage from sendDocument, getFile, admin operations.

Outcome

OPP's existing exfiltration and write-bypass queries would automatically cover these endpoints once modeled. Should be able to answer questions like:

• Can this bot upload files to Slack, or only post messages?
• Can the agent read DMs?
• Can data from GitHub reach Telegram through an unintended binary path?

Context

Motivated by social-monitoring deployment review. Spike -- validate against specific security questions from that deployment. Tracks under OS-43.

Read the existing API descriptor and credential registry patterns under crates/openshell-prover/registry/. Add corresponding Slack, Discord, and Telegram models that distinguish the read, write, upload, messaging, and administrative actions described in the issue. Validate the models against OPP’s existing exfiltration and write-bypass queries and the deployment security questions listed here.