Enforces the gutt GitHub workflow: commit to feature branches, create PRs, request Copilot review before merge. Use this skill whenever interacting with GitHub repos — creating PRs, pushing code, branching, reviewing, or any git operation against a remote, whether via GitHub MCP or the gh CLI. Triggers on: push…
Handles the gutt-pro release process: version bump, tag creation, GitHub release, and marketplace update. Also use for diagnosing version sync issues. Triggers on: release, bump version, publish, tag, new version, ship it, cut a release, marketplace update, version mismatch, can't update, up to date but wrong version…
Register a memory identity and use it — the protocol for any agent that reads and writes the gutt knowledge graph as itself. Register once, recall your own scope before the group's, and tag every write you make. Builds on memory-search (recall) and memory-capture (write) and layers agent identity on top. Triggers on…
Work out what to do when two memories disagree — both stored, or one stored and one about to be written. Given one specific pair, gather the evidence — who asserted each, what standing that carries, whether the two even apply to the same thing — then recommend exactly one of supersede, coexist, or escalate, or report…
Walk the gutt knowledge graph's relationships to answer what a summary can't — how things connect, what depends on or blocks what, and whether a fact still holds. The deep rung of memory search (rung 3): it runs after a first pass has surfaced the relevant entities, never as the first call. Triggers on: how is X…
Write to the gutt knowledge graph with discipline — the capture counterpart to memory-search. Use when something learned, decided, or agreed is worth persisting for future work; classify it, dedup against what's already there, and write one self-contained episode. Auto-write covers only Insight and Incident — Lesson…
Deprecated alias for the memory-search skill. Superseded during the 3.0 memory curriculum (GP-856). Invoke memory-search instead; this remains only so the /gutt-pro:memory-retrieval command keeps working.
Search organizational memory efficiently before any non-trivial task — an adaptive, relevance-gated first pass over the gutt knowledge graph, deepening only when needed. Use to recall prior decisions, lessons learned, past work, patterns, owners, or why something was done. Triggers on: previous, before, last time…
Move Claude Code's own per-project memory store into the gutt knowledge graph, once, with the user's consent — back it up, write the episodes, verify they landed, then remove the local copies and leave a note that gutt is the store of record. Use when the user accepts an offer to migrate local memory, or asks to move…
First-use onboarding for gutt memory integration. Verifies MCP connection, demonstrates memory search and capture, explains HUD and active hooks. Triggers on: onboard, getting started, first time, setup guide, how does this work.
Shape a reply so the answer is the thing the user is left looking at: substance first, no preamble or closing pleasantry, state restated rather than assumed carried, finished work shown concretely, lists capped and ranked, concrete estimates, one next action last. Defines what the closing summary of a turn contains…
Analyze team workflows to identify missing skills, agent behavior gaps, or repeated corrections that indicate need for new skills. Use when analyzing what skills would help, identifying why agents keep making the same mistakes, conducting skill gap analysis, or when user asks what skills would help, why do agents keep…
Build a cited recap of a time window for a person, project, or team — what happened, what was decided, which incidents and lessons landed, and where the subject was mentioned — from organizational memory, enriched with work-tracker activity when those tools are present. Produces a plain-language, cited report in the…
Turn a bug into an investigation brief — how bad it is and why, which area it most likely lives in, and the similar failures the organization has already paid for, every claim cited. Produces a brief the developer acts on, never a severity, area, or status written into Jira. Use when a bug lands and nobody yet knows…
Review a pull request against what the organization already knows — its recorded standards and agreements, the findings this team has accepted before, and the incident history of the files being touched — with each finding verified at the source before it is reported, and accepted findings offered up as lessons…
Break one story into the sub-tasks it actually decomposes into — each with a title, checkable acceptance criteria, a grounded effort range, and its dependencies — in Jira's own vocabulary and nobody else's. Produces a proposal to argue with, never sub-tasks written into Jira unless explicitly asked for. Use when a…
Check a Jira ticket — or a draft not yet filed — for duplicates and overlapping work before effort is spent twice: candidates from Jira search and from organizational memory, each resolved to a verdict with evidence and confidence. Use when a ticket smells familiar or before filing a new one. Triggers on: duplicate…
Ground a Jira ticket's effort estimate and risk picture in comparable past work — what similar tickets actually involved, what bit the team before, which areas carry incident history. Produces input to the team's estimate, never a number written into Jira. Use when sizing a ticket, preparing refinement, or asking what…
Build the context brief a developer needs before starting a Jira ticket — what the ticket itself says, what organizational memory adds (the decisions, meetings, and people behind it), related past work, and what is still missing. Use before picking up a ticket whose background is thin, scattered, or assumes meetings…
Write a development program record — goals, milestones, check-in cadence — into the individual's personal memory scope as one self-contained episode a later session can pick up cold. This is the record-shaping half the mentor and onboarding-guide agents run after grounding a request in org memory — a fresh ask to grow…
Read a person's development program and its check-in history out of personal memory, report where they actually stand, and write the next check-in chained to the last one — so a fresh session reconstructs goals, milestone status, and open questions without the user re-explaining any of it. Use for check-ins, progress…
Scan a JQL-scoped slice of a Jira backlog for tickets that are really the same work: duplicate and overlap clusters with cited evidence, consolidation proposals that map source tickets to one drafted item, and stale candidates each carrying its justification. Propose-only — every close, cancel, merge, or link waits…
Rank a slice of the Jira backlog on the organization's own criteria, with each item's position justified by cited evidence — what was decided, what was promised to a client, which areas keep breaking — and close with a one-page summary for leadership. Produces a proposal to argue with, never a rank written into Jira.…
Draft Jira-ready stories from source material — a meeting transcript, a wiki page, a freeform ask — and manage the ones already filed: structured updates, splits into sibling stories, links, and refreshes of stale text. Every draft cites its source and carries testable acceptance criteria; every create and edit is…