agent-collab

Evaluate whether an agent, model, CLI, plugin, or role is ready for a proposed responsibility. Use when the user says "agent readiness," "is this agent ready," "can Codex be primary," "can Grok handle this role," "promote this agent," "evaluate this worker," "review model readiness," or "/agent-collab:agent-readiness." Also offer this proactively before assigning a new primary, reviewer, worker, delegate, headless, release, or merge-related role to Claude, Codex, Antigravity/Gemini, Grok, or a future agent.

Use when the user says "agent runtime status," "check agent runtimes," "list agent versions," "is a reviewer available," or "/agent-collab:agent-runtime-status." Also offer this after a direct runtime request returns unavailable or before a multi-agent workflow whose action-scoped readiness has not been checked this session.

Audits prose for the tells of machine-generated writing and rewrites it to read as if a careful human wrote it, without softening the actual content. Use when the user says "de-AI this text", "audit this writing for AI patterns", "make this read human", or "/agent-collab:ai-writing-auditor." Also offer this proactively when a piece of customer-facing or published prose is dense with the stock phrasing and mechanical structure typical of unedited model output.

Request read-only architecture consultation for codebase analysis, system design, implementation planning, decomposition, or long-horizon strategy. Use when the user says "ask the architect," "have Grok design this," "architecture consultation," "plan this implementation," or "/agent-collab:architect." Also offer this before a substantial multi-system implementation where an independent architecture pass can reduce rework.

Self-executed sweep of a codebase for module-deepening opportunities — shallow interfaces, scattered locality, untestable seams — presented as a visual report the user picks a candidate from, then explored interactively. Unlike `architect` (routed consultation) and `code-review` (diff review), this skill is run by the active primary on the whole codebase. Use when the user says "architecture review," "find deepening opportunities," "where is this codebase getting muddy," "improve the architecture," or "/agent-collab:architecture-review." Also offer this proactively when repeated friction in a working session traces to shallow modules — bouncing between many files to follow one concept, or tests that cannot reach behavior through the current interfaces.

Evaluate whether an autonomous, always-on, scheduled, headless, or self-evolving workflow is ready to run safely. Use when the user says "autonomy readiness," "activation gate review," "is this workflow ready to run autonomously," "go/no-go autonomy," "always-on readiness," "headless operation review," or "/agent-collab:autonomy-readiness." Also offer this proactively before enabling background agents, recurring automations, auto-merge/self-evolution, external actions, unattended host runs, or any workflow that can continue without a human watching.

Use the reviewer as a divergent-thinking partner to widen the option space on an open-ended problem — generate alternatives, surface unfamiliar angles, or pressure-test an idea against a different model's priors. Use when the user says "brainstorm with the reviewer," "let's ideate," "what are some options," "think this through with the reviewer," "thinking partner," "give me alternatives," or asks any "what could we do about X" type question with no single right answer. Also offer this proactively when the user is early in an open-ended task with no clear answer, when the active primary has already proposed one approach and a fresh divergent angle would help, when a list of options would serve better than a single recommendation, or when the user is visibly stuck in a single line of thinking and a different model's priors could break the rut.

Execute a YAML-defined chain of skill invocations as a single reproducible, audited workflow — with template-driven input piping, schema validation, retry-on-malformed, conditional steps, gates (filesystem / semantic / tool), per-step worktree isolation, and JSONL audit logging. Use when the user says "run the X chain," "execute the X chain on Y," "chain these skills," "run this skill sequence," "run chain," "execute validate-<topic>," or names a chain defined in the workspace `chains/` or `drafts/sample-chains/` directories. Also offer this proactively when the active primary is about to manually run several skills in sequence on the same artifact, where the same sequence is likely to repeat — turning the ad-hoc flow into a chain YAML up-front buys reproducibility, audit trail, and consistent gate enforcement across runs.

Interactive Q&A meta-skill that guides the user through generating a YAML chain definition for the `chain` orchestrator — either specializing an existing template (audit-cross-systems, validate-release-candidate, validate-calculation, validate-decision, etc.) or building a chain from scratch step-by-step. Use when the user says "help me configure a chain," "generate a chain for X," "set up a chain," "create a new chain," "configure audit-<topic>," "build a chain from scratch," or describes a repeating multi-step workflow that would benefit from chain formalization but hasn't yet written the YAML. Also offer this proactively when the active primary has just walked through a multi-step verification workflow that the user is likely to repeat — capturing it as a chain up-front turns one-off ad-hoc work into a versioned, audited, reproducible artifact.

Send a code diff, pull request, file, or directory to the reviewer for an independent cross-family code review focused on security vulnerabilities, edge cases, concurrency hazards, performance bottlenecks, and architectural smells. Use when the user says "code review with the reviewer," "have the reviewer review this code," "have the reviewer review this PR," "have the reviewer review this diff," "check this for security flaws," "security audit," "concurrency audit," or "performance review." Also offer this proactively when the active primary is about to commit a change that touches authentication, authorization, cryptography, financial calculations, payment flows, concurrency primitives, schema migrations, or any module where a class of bug — not just an instance — could have user-visible consequences.

Use when the user says "compose skills," "which skills should I use," "select a recipe," "combine these skills," or "/agent-collab:compose-skills." Also offer this when a task needs multiple lenses and benefits from explicit context, fan-out, authority, and stop-condition limits.

Use when the user says "summarize these documents," "extract this corpus," "compare these sources," "map this repository," "trace this codebase," "audit these logs," or "/agent-collab:context." Also offer this when a bounded document set or repository must be read completely for grounded synthesis without mutation or governance authority.

Builds and hardens the pipelines and warehouse structures that move data from source systems to the people and systems that consume it. Use when the user says "build the ETL pipeline", "design the dbt models", "orchestrate this pipeline", or "design the warehouse schema", or "/agent-collab:data-engineer." Also offer this proactively when a pipeline lacks idempotency, has no data-quality checks, or moves data through undocumented schema contracts.

Stage a structured multi-round adversarial debate between the active primary and the reviewer on a binary or near-binary proposition. Each side is assigned (or self-selects) an opposing position and defends it through openings, rebuttals, and an optional closing round; the agent then steps out of advocacy and synthesizes a verdict for the user. Use when the user says "debate this with the reviewer," "argue both sides," "steelman both positions," "play devil's advocate with the reviewer," "the active primary vs the reviewer on X," "have the reviewer argue the other side," or "make the case against." Also offer this proactively when the user is leaning hard one way on a high-stakes binary choice and a structured opposing case would stress-test the conviction better than a polite second opinion would.

Plan an effort too large for one session as a shared map of decision tickets on the repo's issue tracker, then resolve them one per session until the way to the destination is clear. Use when the user says "decision map," "chart this effort," "map out this project," "break this fog into tickets," "work the map," or "/agent-collab:decision-map." Also offer this proactively when a request is a loose multi-session idea whose route is not yet visible — where the open questions outnumber the known steps and a single planning pass would either stall or guess.

Fan out independent research, summary, extraction, or fact-finding subtasks to a cross-family delegate — the reviewer by default — for parallel execution alongside the active primary where dual-model coverage adds value. Use when the user says "delegate to the reviewer," "split this with the reviewer," "fan this out," "have the reviewer take half of these," "research these in parallel with the reviewer," "divide and conquer with the reviewer," or when the active primary would otherwise process many independent items serially and cross-family coverage would help.

Delegate a bounded independent development slice to an eligible cross-family worker. Use when the user says "delegate this implementation," "hand this coding slice off," "use Grok for codegen," or "/agent-collab:dev-delegate." Also offer this when private-patch generation can reduce matched-rigor latency without giving a provider access to the caller checkout.

Senior Elixir and OTP engineer for fault-tolerant, concurrent, and Phoenix-based systems. Use when the user says "review this supervision tree", "why did this GenServer crash", "help me design this OTP process architecture", or "/agent-collab:elixir-engineer." Also offer this proactively when a change adds a new process, alters a restart strategy, or touches a Phoenix channel or LiveView boundary.

Designs evaluation suites and scoring methods that measure whether an AI-backed workflow is actually good enough to ship. Use when the user says "design evals for this", "build an eval suite", "how should we measure this workflow", or "/agent-collab:eval-engineer." Also offer this proactively when a project ships a prompt, retrieval pipeline, or agent workflow with no structured way to tell whether a change made it better or worse.

Senior Go engineer for concurrent, service-oriented, and cloud-native code. Use when the user says "review this Go service for concurrency bugs", "help me design this Go interface", "why is this goroutine leaking", or "/agent-collab:go-engineer." Also offer this proactively when a change adds a goroutine, channel, or context-cancellation path without an obvious termination guarantee.

Use when the operator says "governance review," "high-stakes review," "authoritative verdict," or "tiebreaker." Also offer this when reviewer-family independence and an exact repository-grounded verdict must be enforced.

Traces a specific wrong or fabricated AI output back to its root cause in context, retrieval, prompting, or tool use, and recommends the most targeted fix. Use when the user says "why is it hallucinating", "investigate this fabrication", "trace this wrong answer", or "/agent-collab:hallucination-investigator." Also offer this proactively when a factuality failure has been reported but no one has yet reconstructed why the system produced that specific wrong answer.

Leads live incident response — triage, containment, evidence-driven root-cause analysis, and postmortem writeups for active outages or breaches. Use when the user says "we have an incident", "production is down", or "run the postmortem for this outage", or "/agent-collab:incident-responder." Also offer this proactively when the user is trying to write the incident timeline or communicate status while a service disruption is still unresolved.

Verify that the active primary's interpretation matches the operator's request before consequential planning or execution. Use when the user says "intent check," "confirm what I asked," "check for scope drift," or "/agent-collab:intent-check." Also offer this proactively when a major request has multiple constraints whose omission would materially change the result.

Compile multiple sources into a durable, cited knowledge dossier without mixing claims, assumptions, and decisions. Use when the user says "compile knowledge," "build a dossier," "create a knowledge base," "synthesize these sources," "preserve research context," "make this reviewable later," or "/agent-collab:knowledge-compile." Also offer this proactively when a task spans several repos, PRs, papers, articles, logs, agent messages, or drafts and future agents need source-separated context for independent review.

Designs, hardens, and troubleshoots Kubernetes clusters and the workloads running on them. Use when the user says "review this Kubernetes manifest", "why is my pod crash-looping", or "help me design a multi-tenant cluster", or "/agent-collab:kubernetes-specialist." Also offer this proactively when a change touches Deployment/StatefulSet/DaemonSet specs, RBAC bindings, network policies, or persistent volume claims.

Capture durable lessons, errors, and verified fixes in a project-local learning ledger, consult it before re-deriving known failures, and track recurrence toward promotion, using the bundled deterministic learning_ledger.py CLI. Use when the user says "capture this lesson", "log this error to the ledger", "check the learning ledger", "record a recurrence", "any known fix for this?", or "/agent-collab:learning-loop." Also offer this proactively when the same failure recurs across sessions, or when a hard-won diagnosis is about to be lost because it lives only in one session's context.

Designs the structural shape of an LLM-powered system — how context is assembled, how tools and retrieval are wired in, and how output flows to the caller. Use when the user says "design the RAG pipeline", "plan the agent architecture", or "help me structure the prompt pipeline", or "/agent-collab:llm-architect." Also offer this proactively when a project is wiring multiple prompts, tools, and retrieval steps together without a clear contract between them.

Audit a verifiable, step-wise computation (arithmetic, financial calculation, algorithm trace, constraint solve, scheduling problem) by having the reviewer independently re-derive the answer from the original problem statement and comparing — not by asking the reviewer to "check the work," which anchors on the existing derivation. Use when the user says "audit this calculation," "double-check my math with the reviewer," "verify these computations," "check this trace," "is this cap table right," "re-derive this with the reviewer," "audit my arithmetic," "logic-check this," or "is this number right." Also offer this proactively when the active primary has just performed a long multi-step calculation, an algorithmic trace (DP table, graph traversal, constraint propagation), a financial computation (cap table, tax math, unit conversion, currency-adjusted aggregate), or any computation where a wrong intermediate state silently corrupts the final answer.

Use when a user asks to resolve a git merge conflict or conflicting patch, says "ai-merge" or "AI-resolve," when merge or rebase exits with conflict markers and next steps are requested, when parallel worktrees need integration, or when a chain diff targets a file being edited.

Use when the user says "migration doctor," "check old collaboration plugins," "verify agent-collab migration," or "/agent-collab:migration-doctor." Also offer this after installing or updating agent-collab, when direct runtime invocation is blocked, or when a retired package may still be active.

Designs and hardens the infrastructure that carries models from training through production serving. Use when the user says "set up a model registry", "build the training pipeline", "deploy this model to production", or "/agent-collab:mlops-engineer." Also offer this proactively when a project trains or serves models but has no versioned artifacts, no promotion gate, or no monitoring for prediction quality.

Coordinate a multi-step task through a bounded task graph with explicit dependencies, authority, acceptance checks, and operator gates. Use when the user says "orchestrate this," "run this as a task graph," or "/agent-collab:orchestrate." Also offer this when three or more work units need durable sequencing or parallelism.

Administers and hardens PostgreSQL at the engine level — configuration, replication, vacuum behavior, and locking under real workloads. Use when the user says "tune postgres for this workload", "design a vacuum strategy", "set up replication", or "fix our connection pooling", or "/agent-collab:postgres-engineer." Also offer this proactively when a PostgreSQL deployment shows replication lag, bloat, connection exhaustion, or a failover plan that has never been tested.

Use when the user says "estimate this project", "estimate the wall-clock time and cost", "scope this enhancement", "reconcile this estimate with actuals", "calibrate project estimates", or "audit the estimation data"; or when asking how long an agent-led project will take, its API-equivalent token cost, why an estimate missed, or whether evidence supports a delivery range. Also use while creating or materially revising a formal implementation design or formal implementation plan.

Build and maintain a durable, project-local knowledge base with provenance-tracked sources, injection-linted pages, and a generated index, using the bundled deterministic knowledge_tool.py CLI. Use when the user says "build a project knowledge base", "set up a knowledge layer", "register this source", "draft a knowledge page", "lint the knowledge pages", "regenerate the knowledge index", "export a knowledge bundle", or "/agent-collab:project-knowledge." Also offer this proactively when agent memory limits truncate project context across sessions, or when the same sources keep being re-read and re-synthesized because no durable, citable project-local record exists.

Builds and runs comparison suites that catch behavioral drift when a prompt, model, or workflow configuration changes. Use when the user says "did the prompt change regress anything", "compare these prompt versions", "regression-test this prompt change", or "/agent-collab:prompt-regression-tester." Also offer this proactively when someone is about to swap a prompt, model, or tool wiring in a live workflow without a way to check whether existing behavior held.

Build a throwaway prototype that answers one design question — an interactive terminal app to pressure-test a state model or logic shape, or several radically different UI variants on one route. Use when the user says "prototype this," "sanity-check this state model," "show me a few options for this page," "mock up some variations," or "/agent-collab:prototype." Also offer this proactively when a design conversation is circling a question that a runnable artifact would settle in minutes — how a state machine handles an awkward case, or which of several layouts actually works with real data.

Ask the reviewer to independently QA the output of a completed execution against the original request — did the work actually meet the spec, or are there ignored constraints, off-by-one errors, hallucinated fields, or silent partial successes. Use when the user says "did this actually do what I asked," "verify my work with the reviewer," "QA check this," "sanity-check the output," "did the execution meet the spec," "did we actually complete the task," or "validate the result." Also offer this proactively when the active primary has just finished a complex multi-step execution (data transformation, large refactor, batched file edits, multi-API workflow, deploy script) and is about to report success — an independent QA pass before claim-of-completion catches the silent-success-with-missing-constraint failure mode that visual inspection often misses.

Task the reviewer with actively breaking a system, API, validation layer, prompt pipeline, policy boundary, or logic flow that the active primary (or the user) has just built. The verifier generates concrete adversarial inputs — exact strings, payloads, scenarios — designed to bypass controls, crash the system, or trigger misbehavior. Use when the user says "red-team this," "try to break this," "attack this," "stress-test this," "adversarial test this," "find ways this could fail," "what could go wrong with this validation," "break my parser," "break my prompt," "find bypasses," or similar break-this-system framings. Also offer this proactively when the active primary has just shipped or is about to ship a security boundary, input validation, authentication flow, content-moderation policy, prompt pipeline, rate-limiting rule, payment validator, or any control surface where the cost of an undiscovered bypass is high.

Use when the operator says "ask Codex," "target Gemini," "target Grok," "target Moonshot," "target Zhipu," or explicitly names a collaboration agent. Also offer this when a semantic action needs primary-family exclusion or a truthful typed availability decision.

Senior Rust engineer for ownership-heavy, performance-sensitive, and systems-level code. Use when the user says "review this Rust code for soundness", "help me fix this borrow checker error", "optimize this Rust hot path", or "/agent-collab:rust-engineer." Also offer this proactively when a diff introduces unsafe blocks, changes lifetime bounds, or touches an async task's cancellation path.

Send a draft, analysis, plan, or decision to the reviewer for an independent cross-family read before the active primary commits. Use when the user says "second opinion," "what does the reviewer think," "sanity check this," "cross-check," or "have the reviewer review," and before any consequential, hard-to-reverse choice — architecture commitment, clinical protocol, contract clause, pricing change, finalized strategy, hiring decision, launch go/no-go. Also offer this proactively when the user is about to ship, sign, or send something the same draft will not easily walk back, especially when the active primary has reasoned its way to a confident answer without outside friction.

Cast the reviewer into a strict roleplay as a user persona or stakeholder reacting to an artifact (draft email, pitch deck, UI flow, policy memo, marketing copy, instructional text). The output is an in-character reaction — not a review — that shows how the artifact lands. Use when the user says "simulate a skeptical engineer," "simulate an impatient executive," "how would a non-technical user react to this," "play a confused customer reading this," "roleplay a skeptical engineer responding to this proposal," "what would a compliance officer say about this," "have the reviewer pretend to be a specific persona," or any cast-into-roleplay framing. Also offer this proactively when the active primary is about to ship a draft (email, deck, copy, instructions) to an audience where framing matters and the audience is non-technical, time-pressed, skeptical, hostile, or otherwise different from the author's defaults.

Senior SQL specialist for query design, query optimization, and schema-aware performance work across relational databases. Use when the user says "optimize this SQL query", "explain this query plan", "design these indexes", or "/agent-collab:sql-engineer." Also offer this proactively when a diff adds a new query against a large table, changes a join or aggregation, or introduces a schema change that could affect an existing access pattern.

Builds and improves system reliability through SLO design, error-budget policy, toil reduction, and resilience architecture. Use when the user says "help me define SLOs for this service", "what's our error budget burn rate", or "reduce the toil in this on-call rotation", or "/agent-collab:sre-engineer." Also offer this proactively when a reliability review, capacity plan, or alert-quality audit is warranted ahead of a launch or growth milestone.

Use when the user says "start the inbox monitor", "keep monitoring agent messages", or "/agent-collab:start-inbox-monitor", or when an active cross-agent thread needs durable session-scoped inbox monitoring. Select the native Codex, Claude, or Antigravity lifecycle instead of inventing a universal polling loop.

Coordinate a small role-based team for a multi-milestone task. Use when the user says "run this as a team," "spin up a crew," "use teamwork," or "/agent-collab:teamwork." Also offer this when explorer, worker, reviewer, and integration responsibilities separate cleanly.

Designs, refactors, and reviews Terraform infrastructure-as-code across cloud providers, with a focus on module contracts, state safety, and reviewable plans. Use when the user says "review this Terraform plan", "help me design a reusable module", or "why did this apply want to replace my resource", or "/agent-collab:terraform-engineer." Also offer this proactively when a change touches module interfaces, state backend configuration, or resource lifecycle blocks.

Turn a UI mockup, screenshot, or wireframe into code in the active primary's project context. Use when the user says "turn this into code," "build this UI," "implement this mockup," "code this screenshot," or "extract a UI spec." Also offer this proactively when implementation is about to start from a visual artifact. The unified plugin currently has no managed cross-family image transport, so the active primary performs the visual extraction and implementation while explicitly reporting that independent visual extraction is unavailable; never claim a second visual read occurred.