bajzi

Plugin bajzi from marketplace bajzi-plugins in bajzaa975/claude-alapcsomag. Commands read from the repo's manifest — 3 days ago.

Skills (20)

• autopilot

  Autonomous work session — carrying several tasks or a sprint through without questions, with a decision log and a closing report. Use it when the user says "autopilot", "I'm not at the machine", "do it all without asking", "work through the list on your own".

• critic

  Run the read-only plan critic on a final execution plan and triage its failure scenarios. Use it when the user says "/bajzi:critic <plan-stem>", or before executing an M/L plan.

• debt

  Parked review debt (runtime/findings/debt.md) - --check the cap before a slice, --drain it with one fixer pass plus one review at a milestone, --calibrate it with a blind reviewer re-rate. Use it when the user says "/bajzi:debt", "check the debt", "drain the debt", "calibrate the debt".

• explain

  Re-explains the last answer, or a named point of it, simply and in the user's language.

• fix

  Fix a round-1 findings file through the bajzi fixer agent (severity-blind copy), run the gate, commit, then run the round-2 re-review and STOP. Use it when the user says "/bajzi:fix <findings-file>", "fix the findings", or /bajzi:review round 1 returned findings.

• grill-me

  A relentless interview to sharpen a plan or design.

• grilling

  Grill the user relentlessly about a plan, decision, or idea. Use when the user wants to stress-test their thinking, or uses any 'grill' trigger phrases.

• handoff

  Save session state into runtime/handoff/<task>.md + generate a suggested opening prompt for continuing after /clear. Use it when the user closes the session, asks for a handoff, is running out of context, or says "handoff", "let's save where we are", "/clear is coming".

• implement

  Implement one slice from runtime/slices/<id>.md through the bajzi implementer agents (Tier 1 -> implementer-risk, else implementer; both sonnet), then commit it for review. Use it when the user says "/bajzi:implement <id>", "implement slice <id>", or a plan step names a slice file.

• level

  Owner-only saver level command, handled by a hook: /bajzi:level [0-3|L0-L3|claude|light|glm|tight] [session|repo|global] | unset [scope]; no argument = status.

• mode

  Switch between day-run (autonomous, model-routed) and normal working mode, or report which one is active. Use it when the user says "mode", "day-run", "normal mode", "which mode am I in", or asks to turn day-run on or off.

• modszertan

  Select or change the repo's leading development methodology (GSD vs superpowers vs neither). Use it when the user says "modszertan", "methodology", "which one do we use", "let's switch to GSD", "don't do the .planning thing", or when the SessionStart hook reported that there is no decision yet in th

• night-run

  Plan an unattended multi-hour overnight autopilot run for one project — preflight, queue, rendered brief, approval gate, and the launch itself, which the session runs once the owner says go. Use it when the user says "night run", "/bajzi:night-run", "overnight autopilot", "plan tonight's run", "run

• project-setup

  Apply or check this repo's project profile (.claude/project-profile.json), incl. start assessment, project size and hosting — project-scope plugins, project .mcp.json entries, .claude/METHODOLOGY, linked skills, instruction files and the bajzi pre-commit gate. Use it when the user says "project-setu

• radar

  Biweekly read-only review of the owner's Claude setup (plugins and their supply chain, Anthropic/Claude Code news, methodology, the owner's own usage) - present the newest radar report and decide which items to adopt, run it now, or install the schedule. Use it on "/bajzi:radar", "/bajzi:radar now",

• respec-cards

  Use when the owner wants a system specified (or re-specified) in their own words as approval cards — "spec cards", "re-spec", "interview me about X", a new project idea before any code, or an existing project whose plan, code and owner intent have drifted apart.

• review

  Review a slice's commit range through the bajzi reviewer agent, write and validate the findings file, and on round 2 apply the close policy (owner / debt) and the debt cap. Use it when the user says "/bajzi:review <slice> <range>", "review slice <id>", or /bajzi:fix calls round 2.

• setup

  Set up a Claude Code machine to the desired state, or only check it for drift — plugin cleanup, then marketplaces, plugins, rtk, settings, the bajzi status line and user-scope MCPs per the manifest. Use it ON A NEW MACHINE, when an existing installation needs tidying up ("set up this machine", "clea

• tired

  Tired mode: all replies in the user's language, plain and short, until it is turned off.

• which

  Answers which installed skill or slash command fits what the owner wants to do, with the exact command to type and where it is installed. Use it on "/bajzi:which <intent>", "which skill", "which command should I use", "what was the command for", "what is the command", "which skill do I need".

Subagents (5)

• fixer

  Fixes every finding in a fixer copy of a findings file in one pass. Tier-agnostic.

• implementer-risk

  Implements one Tier 1 slice.

• implementer

  Implements one specified slice, tests included. Tier 2/3 slices only.

• plan-critic

  Reads a final execution plan, its slice files and the code, and returns ranked failure scenarios as the critic file. Read-only.

• reviewer

  Reviews a commit range and returns a findings file as its final message. Read-only. Never fixes.

Hooks (20)

• SessionStart

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/handoff-load.sh"

• SessionStart

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/methodology-guard.sh"

• SessionStart

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/day-run-mode.sh"

• SessionStart

  node "${CLAUDE_PLUGIN_ROOT}/skills/radar/radar.js" notice

• SessionStart

  node "${CLAUDE_PLUGIN_ROOT}/skills/setup/install-statusline.js" refresh

• SessionStart

  node "${CLAUDE_PLUGIN_ROOT}/hooks/node/writer-guard.js" notice

• SessionStart

  node "${CLAUDE_PLUGIN_ROOT}/hooks/node/tired.js"

• SessionStart

  node "${CLAUDE_PLUGIN_ROOT}/hooks/node/ask-owner.js"

• SessionStart

  node "${CLAUDE_PLUGIN_ROOT}/hooks/node/session-signal.js"

• PreToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/noise-filter.sh"

• PreToolUse

  node "${CLAUDE_PLUGIN_ROOT}/hooks/node/pre-tool.js"

• PreToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/dispatch-guard.sh"

• PostToolUse

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/routing-counter.sh"

• PostToolUse

  node "${CLAUDE_PLUGIN_ROOT}/hooks/node/post-tool.js"

• UserPromptSubmit

  node "${CLAUDE_PLUGIN_ROOT}/hooks/node/session-signal.js"

• UserPromptSubmit

  node "${CLAUDE_PLUGIN_ROOT}/hooks/node/level.js"

• Notification

  node "${CLAUDE_PLUGIN_ROOT}/hooks/node/session-signal.js"

• Stop

  node "${CLAUDE_PLUGIN_ROOT}/hooks/node/session-signal.js"

• StopFailure

  node "${CLAUDE_PLUGIN_ROOT}/hooks/node/session-signal.js"

• SessionEnd

  node "${CLAUDE_PLUGIN_ROOT}/hooks/node/session-signal.js"

Similar plugins