manager

Plugin manager from marketplace greenblacked-ai in greenblacked/AI. Commands read from the repo's manifest — 2 days ago.

Skills (22)

• ai-enablement

  Assess how a team uses AI-assisted engineering and produce a rollout plan — baseline current usage, measure the signals of value while refusing the ones that create compliance theatre, check readiness gates on data handling, licensing, review accountability and cost, place the team on the enablement

• compensation-benchmarking

  Build or repair the pay structure a team is managed against — level fixed before price, sources matched to stage and market, bands with a midpoint and a spread, compa-ratio and range placement, compression, a pay-equity check by cohort, offer construction, and a merit budget against a written matrix

• decision-record

  Produce architecture decision records (MADR 4.x ADRs) and longer design docs / RFCs, choosing between them by stakes and reversibility, and enforcing the structure that makes a decision auditable a year later: labelled artefact type, at least two genuinely considered options with the reason each was

• delivery-review

  Assess the delivery health of a team or programme and say what to change — DORA metrics read the way they were defined, flow and capacity reality, prioritisation frameworks matched to the decision actually being made, and OKRs graded as designed. Produces a ranked findings report with impact and eff

• design-team-cadence

  Design or repair the recurring operating cadence of an engineering team, platform group, or organisation: one-to-ones, staff meetings, planning and operational reviews, architecture forums, skip-levels, retrospectives, and escalation paths. Produces a minimal meeting and decision system with explici

• difficult-conversation

  Prepare and hold a conversation you keep putting off — underperformance, a boundary, an unpopular decision, a conflict between two people. Fix the outcome, separate behaviour observed from the story about intent, say whether it is feedback or a decision already made, write the opening two sentences

• growth-review

  Prepare a performance or growth conversation from evidence rather than recency — a written review, a promotion case for a report, a levelling or calibration argument, or a development plan. Enforces behaviour, impact and evidence for every claim, levelling against the ladder in use, no-surprise deli

• headcount-case

  Build the written case for headcount, a budget increase or a backfill that a finance partner or executive will approve. Opens with the ask in one line (roles, level, loaded cost over 12 and 24 months, date) and the decision required, prices the cost of not hiring from the team's own data, beats two

• hiring-loop

  Design and run a structured hiring loop for an engineering role — the scorecard first, then stages mapped to competencies, questions and written rubrics, work samples that respect the candidate's time, calibrated interviewers, and a debrief that decides on evidence. Use this skill whenever the user

• incident-comms

  Communicate outward during and after an incident — status page posts, customer notices, executive briefings and internal channel updates. Covers a dedicated comms role, acknowledging on impact rather than on diagnosis, a next-update promise you keep, severity-to-comms rules agreed in advance, regula

• inherited-team-audit

  Run the first 30 days of taking over an existing team or a departing manager's reports — change nothing structural and say so on day one, hold a 1:1 with every report using the same five written questions, read two quarters of postmortems, review comments and plans, sit with upstream and downstream

• okr-planning

  Set goals for a team or organisation that change what people do, then review them honestly at the end — objectives as outcomes, key results as evidence with a baseline before a target, guardrails that must not degrade, a small defended number of them, confidence check-ins, and scoring whose real out

• onboarding-plan

  Plan and run the first ninety days for a new engineer — access working before day one, a briefed buddy, a merged change in week one, 30/60/90 expectations as capabilities, the context that is not in the code, and check-ins that capture confusion while it is still visible. Use this skill whenever som

• oncall-health-review

  Assess whether an on-call rotation is sustainable for the people on it and fix the structure — 90 days of paging data per person (out-of-hours pages, sleep-interrupting pages 23:00-07:00, pages per shift, time to resolve), flags for any night with more than two interrupting incidents, rotation size

• performance-improvement-plan

  Run the formal performance-improvement process after the informal conversation has not worked — deciding plan, role change or managed exit, writing a plan whose gaps are observable behaviour with the support offered, the measures and a dated timeline, partnering with HR and legal from day one, runni

• postmortem

  Write or review a blameless incident postmortem in the Google SRE shape — numbered title, quantified impact, root causes separated from the trigger, how it was detected, and action items that each carry an owner, due date, priority and tracker link. Use this skill whenever an incident needs writing

• reorg-design

  Design, sequence and announce a reorganisation or team-boundary change — name the problem current boundaries cause with evidence (cross-team handoffs per project, who edits whose code, ticket wait time) before drawing boxes, design against the work and assign people in a separate pass, test each can

• retention-review

  Review why people actually leave and what keeps them — regretted-attrition analysis over time rather than one exit interview, themes aggregated across departures, stay interviews that surface a risk before the resignation, flight-risk signals, engagement-survey results turned into a few owned action

• status-update

  Write weekly team status, executive updates and stakeholder comms that lead with the answer — BLUF (status flag, bottom line, ask, by when, since last, next, risks) for reporting, SCQA for persuasion, and never the two mixed. Enforces the discipline that makes an update trustworthy: no invented numb

• team-skills-matrix

  Find single points of knowledge on a team and plan cross-training. Rows are capabilities taken from the last six months of incidents and the next two quarters of roadmap, scores each person zero to three by observable behaviour, flags any capability with fewer than two people at level two or above,

• tech-debt-case

  Turn technical debt into a prioritised, defensible investment that product and executives will fund. Prices each item as interest (recurring engineer-hours a month from ticket and incident history, plus probability times impact) against principal (the one-off effort to remove it), ranks by interest

• vendor-evaluation

  Run a buy decision to a defensible conclusion, including the decision not to buy: start from a written requirement and the current cost of the problem, do an honest build-versus-buy that counts three years of operating cost, agree scorecard weights before any demo, run a time-boxed proof of concept

Slash commands (2)

• /manager:agent-ready

  Clone a repository and run every fenced shell command its AGENTS.md contains — on this machine, with a stripped environment rather than a sandbox — after listing the commands and stopping for confirmation. Use only on a repository you would already trust enough to run its test suite. Also checks ins

• /manager:weekly

  Draft a weekly status update from what actually shipped — merged pull requests, commits and closed issues — bottom line first, with every number sourced or marked missing.

Subagents (2)

• contract-reader

  Read a vendor contract, order form, DPA, SOC 2 report or security questionnaire and return the clauses that decide the deal — pricing metric and scaling, overage, renewal uplift and notice, termination, data ownership and export, sub-processors, residency, breach notification and exit assistance — e

• feedback-synthesiser

  Read collected peer and stakeholder feedback about one person and return themes, each with the number of independent sources and a quoted example, separating observed behaviour and impact from unusable statements about personality. Use when written feedback has been gathered for a growth conversatio

Similar plugins