sixth-sense

Archive a session's handoff into a searchable, git-committed project history, for the subset of sessions worth keeping beyond `handing-off`'s ephemeral temp file. Explicit invocation only.

One-time interactive setup — configures model preferences and effort levels for Sixth Sense subagents.

Use before applying test-driven-development, diagnosing-bugs, or requesting-code-review to detect whether the task targets Java/Spring Boot backend, Angular, React, Next.js, Python, or other stacks, and route to the correct conventions.

Use when encountering bugs, test failures, or unexpected behavior — replaces random fixes with a disciplined, hypothesis-driven diagnosis loop. Trigger on "diagnose", "debug", "broken", "failing", "throwing".

Use when facing multiple independent tasks that can be worked on without shared state or sequential dependencies — dispatches specialized subagents with Effort Routing, batching, and agent cap.

Build and sharpen a project's domain model — pin down terminology, record architectural decisions, maintain CONTEXT.md and ADRs.

The nightmare investigator — hunts edge cases and runs premortems on failure modes. Invoke it directly with /investigate, or it gets called by requesting-code-review, implementing-work, synthesizing-spec, and domain-modeling.

Use when implementation is complete, all tests pass, and you need to decide how to integrate the work — guides completion of development work by presenting structured options for merge, PR, or cleanup.

Produce a compact handoff document for a long or interrupted session, so a fresh session (or the same one after context compaction) inherits the live thread instead of re-deriving it. Explicit invocation only.

Implement the work described by the user in the spec or tickets — orchestrates TDD, typechecking, code review, and commit.

Interview the user to turn a rough request into a polished, portable prompt for an external tool or model — not for building here. Use when the target isn't this session (another LLM, an image generator, a different agent) and the ask is "write me a prompt for X" / "help me phrase this for Y" / "come lo chiedo a [tool]".

Use when completing tasks, implementing features, or before merging — dispatches Dylan Dog (edge-case hunting) and processes feedback with technical rigor, not performative agreement.

Resolve an in-progress git merge or rebase conflict by understanding both sides' intent, not by guessing or aborting. Use when `git status` shows unmerged paths, a merge/rebase is stuck on conflicts, or the user says "merge conflict"/"rebase conflict".

Put the user's plan or design through a relentless round of questioning until a shared understanding is reached. Use when the user wants to pressure-test a plan or design before building.

Turn the current conversation into a structured spec document — no interview, just synthesis of what has already been discussed.

Use when implementing a feature or fixing a bug. Enforces red-green-refactor. Loads stack-specific conventions for Java/Spring Boot, Angular, React/Next.js, or Python.

Creates isolated git worktrees with systematic safety checks and project setup automation. Use when starting feature work that needs isolation from current branch, before executing tasks with subagent-driven development, or anytime you need an isolated workspace for a new branch.

Use when starting any conversation — establishes how to find and use skills, and platform adaptation.

Use when about to claim work is complete, fixed, or passing, before committing or creating PRs — requires running verification commands and confirming output before making any success claims; evidence before assertions always.