AI Workflows

Operational workflows for AI development

Documentation, planning, context management, and communication systems for teams building with AI tools.

9 workflows online

contextonline

Brain

Project memory, retrieval, and session hygiene for AI development work.

Use Brain when repo knowledge needs to survive beyond one chat: architecture notes, workflow rules, implementation decisions, risks, and focused context packets.

Commands

brain prep --task "..."brain search "slopstack ..."brain session run -- npm run lint

Best for

  • Loading the right repo context before edits
  • Capturing decisions after meaningful changes
  • Keeping agent sessions auditable
memoryretrievalsessionscontext
Read related guide
planningonline

Plan

Spec-first planning for turning ideas into scoped execution contracts.

Use Plan when an idea needs shaping before implementation: brainstorm, refine, challenge, promote to spec, and execute with traceable scope.

Commands

plan status --project .plan brainstorm start --project . "..."plan spec execute --project . <spec-slug>

Best for

  • Separating discovery from implementation
  • Keeping specs canonical
  • Avoiding vague multi-feature work
specsroadmapbrainstormexecution
Read related guide
communicationonline

Caveman

Token-efficient agent communication that keeps technical signal intact.

Use Caveman when speed and low-token communication matter. It strips filler while preserving exact technical names, commands, risks, and code.

Commands

/caveman lite/caveman full/caveman ultra

Best for

  • Long debugging sessions
  • Low-noise code reviews
  • Fast status updates with exact commands
tokenscommunicationskillsworkflow
Read related guide
documentationonline

Agent Docs

AGENTS.md plus focused docs for reusable project instructions.

A lightweight documentation pattern for AI coding tools: keep the root contract small, then route agents to architecture, content, UI, and task guides only when needed.

Commands

Read AGENTS.mdUpdate docs/ when workflows change

Best for

  • Reducing repeated project orientation
  • Keeping repo conventions discoverable
  • Sharing workflow knowledge across tools
agents.mddocsskillsinstructions
Read related guide
automationonline

Sandboxed Agent Change Loop

A bounded execution loop for agent changes that isolates work, checkpoints progress, gates risky actions, and publishes only verified artifacts.

Use this loop when an agent can edit files, run commands, call external tools, or work across long-running sessions. It separates the control plane from disposable compute, keeps credentials outside the workspace, and makes approval plus verification explicit before anything leaves the sandbox.

Commands

Create one isolated workspace per taskPersist checkpoints outside the sandboxRequire approval before external writesRun project checks before publishing artifacts

Best for

  • Coding agents that modify repositories
  • Computer-use agents operating browsers or desktops
  • Long-running work that must survive retries or restarts
agentssandboxapprovalsdurable-execution
Read related guide
communicationonline

Collaborative Agent Handoff

A conversation-to-PR workflow that turns shared decisions into bounded agent work with explicit targets, provenance, validation, and review.

Use this workflow when a team wants to start or steer an agent from Slack, Teams, or another shared discussion. It filters the thread into a durable brief, separates participation from write authority, pins the repository and branch, and carries the work through an independently reviewed pull request.

Commands

Name the repository, base branch, issue, and write-authorized ownerSummarize the thread into goals, non-goals, decisions, and acceptance criteriaStart one shared agent session for one taskVerify the diff, checks, provenance, and required approvals

Best for

  • Turning product or engineering discussions into implementation work
  • Shared incident follow-ups that need a traceable code change
  • Teams that want visible agent steering without weakening merge controls
collaborative-agentshandoffchatpull-requests
Read related guide
automationonline

Agent Observability Loop

A privacy-aware telemetry loop that connects agent traces and adoption metrics to reviewed outcomes.

Use this loop to instrument production agents without treating transcripts as observability. It defines stable identities, traces workflow and tool boundaries, limits payload capture, emits low-cardinality metrics, and joins each run to evidence such as review, merge, deployment, or rollback status.

Commands

Assign stable agent, workflow, run, task, and trace IDsTrace agent, model, plan, retrieval, and tool operationsRedact sensitive payloads before telemetry exportJoin each run to review, deployment, or business outcomes

Best for

  • Production agent rollout and incident diagnosis
  • Comparing workflow versions without vanity metrics
  • Auditing privacy, cardinality, and trace completeness
agentsobservabilityopentelemetryprivacy
Read related guide
automationonline

Risk-Calibrated Agent Review

A review loop that matches scrutiny to change risk, turns findings into testable claims, and rechecks the final commit before merge.

Use this workflow for agent-authored or AI-reviewed changes that need more than a generic diff pass. It establishes a risk contract, bounds review context, separates blocking findings from suggestions, records resolution evidence, and keeps final acceptance with an independent owner.

Commands

Classify the change by impact, sensitivity, and system boundariesChoose review depth and required independent ownersAdjudicate every finding with evidence and a resolution reasonRerun checks and review the final head commit before merge

Best for

  • Pull requests authored by coding agents or bots
  • Security-sensitive, cross-system, or unusually large changes
  • Teams using automated review without surrendering merge authority
agentscode-reviewpull-requestsrisk
Read related guide
automationonline

AI Model Retirement Migration

A deadline-aware workflow for finding model dependencies, proving replacement behavior, and removing retired endpoints without service interruption.

Use this workflow when a provider or AI platform deprecates a model, alias, or request parameter. It turns the notice into an owned inventory, tests the replacement against an explicit behavior contract, stages a reversible rollout, verifies that retired traffic reaches zero, and removes the old path before shutdown.

Commands

Inventory model IDs, aliases, policies, SDK defaults, and hidden fallbacksRecord the retirement deadline, replacement, owner, and failure modeTest the replacement against behavior, safety, latency, and cost gatesCanary the change with observability and a time-bounded rollback pathProve zero retired-model traffic, then remove compatibility code

Best for

  • Applications pinned to provider model IDs or preview endpoints
  • Enterprise coding assistants governed by model-access policies
  • Agent workflows whose tools, schemas, prompts, or costs change by model
modelsdeprecationmigrationreliability
Read related guide