What problem does it solve? Agents often post long, unstructured comments to GitHub, Linear, Slack, and Notion that bury the key takeaways. This Skill sets up mechanical enforcement of the dual-track comms format (terse top level plus a collapsible or threaded full-context section) so every outbound post is checked before it leaves the machine, without relying on the agent remembering the rule. ## Core Features & Use Cases - Adaptive environment discovery: Detects the user's agent harness (Claude Code, Amp, Codex, OpenCode) and posting tools (gh CLI, curl, MCP servers) before installing the matching interception mechanism. - Reference hook scripts: Bundled bash PreToolUse hooks check GitHub comments/reviews, Linear GraphQL commentCreate calls, Slack chat.postMessage payloads, and Notion MCP comments against length and collapsible-marker rules, with an override token and JSONL telemetry. - Advisory enforcement contract: Denies non-compliant posts with corrective guidance, fails open on unparseable input, and never polices long-form page content. - Use Case: A team lead asks their agent to set up the dual comms hooks; the Skill discovers they use Claude Code with the gh CLI and Notion MCP, copies the scripts to ~/.claude/hooks, registers the matchers in settings.json, and runs the bundled tests to verify allow, deny, and override behavior. ## Quick Start Ask your agent to install and configure the dual-track comms enforcement hooks for your current harness and posting tools.
