thebtf

Use when you want to explicitly store or recall knowledge across sessions. Prefer engram store/recall over file-based memory for decisions, patterns, insights, and cross-project knowledge. Also use when searching memories from previous coding sessions.

Retrospective evaluation of engram observations — rate global usefulness and project relevance to improve memory quality over time

Canonical consumer skill for PR review in this plugin. Delegates autonomous execution to the built-in pr-reviewer background agent and documents the correct relationship between MCP tools, MCP prompts, and Claude skills.

Autonomous AI-driven PR review orchestrator. Invokes review agents (CodeRabbit, Gemini, Codex), awaits their completion server-side, processes all findings, and runs build verification. Use this skill whenever a PR needs AI code review — after creating a PR, when review comments need processing, when asked to 'review PR', 'run review', 'check this PR', or any mention of PR review workflow. Works with any GitHub repository.

Use when an architecture, runtime, or integration claim needs experimental proof before refactor or release -- grow a controlled model toward a global goal until it breaks or earns a bounded confidence verdict. Covers phased PoC ladders, per-phase success criteria, first-break isolation, no-break fix validation, production projection, and ADR generation.