By kucherenko
Execute autonomous multi-agent Heist pipelines for spec-driven feature development: reconnaissance scans codebase, adversarial grilling brainstorms and refines architecture, drafts binding specs, sets up git-isolated worktrees with TDD-enforced agent coding, integrates changes via merges and audits, enforces governance guardrails and institutional memory throughout.
npx claudepluginhub kucherenko/gangsta --plugin gangstaReject a pending-don-confirmation Heist; relocates its directory to docs/gangsta/.aborted/ with marker
Sign and execute the most recent (or named) autonomous Heist's Hit + Laundering phases
Run the autonomous Heist pipeline (Recon -> Grilling -> Sit-Down -> Resource-Development) for a feature; produces signed-by-don-proxy Contract and Plan
Use this agent for specialized external tasks — codebase scanning, dependency audits, documentation retrieval, API integration, and other support work. Associates are not family members but provide essential services.
Use this agent during The Grilling to attack both the initial idea and proposed solutions. In Round 1, challenges the objective itself (problem framing, hidden assumptions, scope). In all rounds, finds architectural flaws, security gaps, constitution violations, and regression risks. Returns REJECT, CHALLENGE, or CONCEDE verdict.
Use this agent during The Grilling to propose architectural solutions based on the reconnaissance dossier. Must include inverse reasoning (self-identified risks) and constitution compliance.
Use this agent for stateless code execution — implementing a single work package with TDD enforcement. Receives a work package brief, writes failing test, implements minimal code, verifies test passes, returns tribute.
Use this agent during The Grilling to mediate between proposer and devils-advocate. Integrates valid attacks and don's feedback into revised solutions. Detects repetition and stronzate.
Use this agent for independent code inspection — reviews git diffs against requirements, categorizes issues by severity, and delivers a verdict on production readiness. Dispatched by the audit-review skill.
Use when all territories report completion after the hit — handles integration, verification, consigliere final review, cleanup, and ledger updates to produce clean production-ready code
Use when enforcing governance guardrails during any gangsta operation — referenced as cross-cutting concern by all skills for anti-hallucination, authorization, state durability, resource management, and spec supremacy rules
Use when receiving code review feedback — processes orders with technical rigor, not blind obedience or performative agreement
Use when beginning a new heist — deploys associates to survey the target codebase, existing tests, dependencies, documentation, and the ledger to produce a reconnaissance dossier for the don's review
Use when the contract is signed and work packages need to be created — decomposes the contract into bite-sized tasks, sets up git isolation, allocates territories and token budgets, producing the execution plan
Use when starting feature work that needs isolation from the current workspace — sets up a secure safehouse with git worktrees for clean operational bases
Use when about to claim work is complete, fixed, or passing — sweeps for evidence before any assertions, because unverified claims are invalid
Use when orchestrating workers within a specific domain territory — managing work package dispatch, reviewing reports against the contract, and reporting status to the underboss
Use when needing impartial architectural advice, security audit, spec integrity review, or a second opinion — operates outside the chain of command with standing authority to invoke truth checks
Use when implementation is complete and all tests pass — guides the clean exit from a development branch with structured options for merge, PR, or cleanup
Use when encountering any bug, test failure, or unexpected behavior — finds the rat in the code through systematic root-cause interrogation before any fix attempts
Invoked only by gangsta:autonomous-mode; substitutes for the human Don across all six Heist phases until /gangsta:go (sign) or /gangsta:abort (reject)
Use when implementing any feature or bugfix — enforces the Red-Green-Refactor drill with no production code allowed without a failing test first
Use when completing tasks, implementing major features, or before merging — dispatches the-inspector to audit the books before the job is closed
Use ONLY when invoked via the `/gangsta:heist` or `/gangsta:go` commands (see `commands/heist.md` and `commands/go.md`) — orchestrates the Heist pipeline autonomously with `gangsta:don-proxy` substituting for the human Don across all six phases until the real Don signs via `/gangsta:go` or rejects via `/gangsta:abort`.
Use when adversarial brainstorming is needed after reconnaissance — runs a multi-agent debate with proposer, devils-advocate, and synthesizer subagents, with the don participating each round, bounded by hard round limits to prevent infinite loops
Use when the execution plan is approved and it is time for parallel execution — dispatches workers through crew leads to implement work packages with TDD enforcement, report collection, and escalation protocols
Use when reading or writing institutional memory — insights from successful solutions, fails from historical mistakes, and the Project Constitution that accumulates commandments and negative constraints across heists
Use when the grilling consensus is approved and it is time to draft the formal contract specification — strictly prohibits code generation, produces the binding spec document that governs all subsequent phases
Use when decomposing tasks into work packages, allocating territories to crew leads, managing heist phase progression, or serving as the operational buffer between the Don and the crews
Use when starting any conversation — establishes how to find and use Gangsta skills, enforces the 1% invocation rule, maps platform tools, and routes the Don's intent through the Gangsta Agents Family hierarchy
Specification-driven development workflow: specify → plan → tasks → implement
Uses power tools
Uses Bash, Write, or Edit tools
Share bugs, ideas, or general feedback.
HelloAGENTS — The orchestration kernel that makes any AI CLI smarter. Adds intelligent routing, quality verification (Ralph Loop), safety guards, and notifications.
AI-powered cascading development framework with design document system and multi-agent collaboration. Breaks down projects into Features (Mega Plan), Features into Stories (Hybrid Ralph), with auto-generated technical design docs, dependency-driven batch execution, Git Worktree isolation, and support for multiple AI agents (Codex, Amp, Aider, etc.).
Long-running agent harness with 5-layer memory architecture, GitHub integration, autonomous batch processing, Agent Teams with ATDD, 9 hooks (safety, quality gates, team coordination), and 6 Agent Skills
OpenAgentsControl — multi-agent orchestration for Claude Code. Context-aware development with skills, subagents, parallel execution, and automated code review.
Describe your goal, approve the spec, then step away — Claude and Codex loop together until it's right.