Add New Element Shape â Agentic Workflow
You are an agent implementing a new element shape in the LikeC4 codebase. Follow this workflow phase by phase. **Do not skip phases or reorder steps.** Pause at marked checkpoints and wait for user confirmation before continuing.
â¡ signal-ckpt â Manual Checkpoint
Collapse the session into a â¤50 token state atom. Drop all prior history. Resume from the atom.
batch-process
Process multiple content pieces through a prioritized, checkpointed queue with progress tracking and per-piece quality gates
mcp-video
Use mcp-video for guarded video editing, FFmpeg operations, media analysis, subtitles, audio workflows, Hyperframes rendering, repurposing packages, and release checkpoints through an MCP server, Python client, or CLI. Trigger when an agent needs to inspect, edit, render, validate, or package local media safely.
Claude Vigil
Checkpoint project state before risky work. Restore if things break.
Auto Create PR (loop)
Turn a free-form brief into an execution plan, implement it **one commit per Step** in an isolated worktree, batch verification proofs at checkpoints, keep a live handoff doc and append-only notification log, and open a labeled PR against the configured base branch.
end
Session-closing skill that helps users wrap up intentionally. Use when: user says done, wrapping up, end my day, closing out, call it a day, goodnight, that's it for today, checkpoint, pause. Bookend to /start. Scans git activity, surfaces what happened, spawns a crystallize agent for deep analysis, commits uncommitted work, and closes with a brief summary. Works for end-of-day, end-of-research-batch, end-of-decision-sprint, or mid-work checkpoints.
Penfield Memory
Persistent memory and knowledge graphs for AI agents. Hybrid search, context checkpoints, and more.
checkpoint-backlog
Checkpoint the ended Claude Code sessions waiting in subrosa's queue. Reads the pending-checkpoint queue and, for each queued past session, extracts durable facts into that project's memory (leaf files + facts database + regenerated MEMORY.md), then clears it from the queue. Run it at session start when there's a backlog; it's also fine to invoke by hand.
Harness â Multi-Agent Orchestration
Orchestrate complex tasks through Planning â Generation ��� Evaluation â Retro. Fresh sub-agents per checkpoint prevent drift. Retro accumulates learning across tasks.
assessment-design
Evidence-based assessment design with rubrics, feedback strategies, and formative checkpoints. Aligns each assessment to learning objectives using Bloom's taxonomy. Applies Nicol's 7 principles of good feedback practice. Reads from /learning-objectives manifest and extends it with assessment specs. (idstack)
80-20-review
Focus code review effort on the 20% of code that causes 80% of issues. Prioritizes data access, security, concurrency, and integration boundaries over formatting and style. Uses blast radius scoring to determine review depth. Includes checkpoint schedules, critical path identification, and a batch review checklist. Load this skill when reviewing code, PRs, or architecture, or when the user mentions "review", "code review", "PR review", "what should I review", "review priorities", "blast radius", or "critical path".
Io.Github.Charo360/Statecli
State replay and self-debugging for AI agents. Track, replay, undo, checkpoint.
SwiftUI Navigation with Navigator
When assisting the user with SwiftUI navigation in a project that uses Navigator or the NavigatorUI library, follow the conventions and patterns below. The skill is agent-agnostic; apply it whenever the user mentions Navigator, NavigatorUI, SwiftUI navigation, deep linking, checkpoints, managed navi
AC Lock Checkpoint
Runs **automatically as Phase 4.0**, after Tasks Validation and before the first implementation task. Purpose: give the user one final explicit confirmation that the Acceptance Criteria represent what they actually want â before any implementation begins.
bounded-loop-runner
Design safe unattended or long-running loops with explicit limits, checkpoints, and handoff artifacts.
design-doc-generate
Generate design documents for Validate mode work items (Checkpoint 1). Required for high-complexity items.
Interview Mode
Structured interviewer for Claude Code — probing questions, decision tracking, evolving checkpoints.
Encode stage -- stop before the loop, compare ALL inputs:
ref_ckpts = {"preloop": Checkpoint(save=True, stop=True)} run_reference_pipeline(ref_ckpts) ref_data = ref_ckpts["preloop"].data
tuios
Drive tuios from inside one of its panes. Find out where you are running, read and write other panes, run work and wait on it instead of polling, report your own state so the person sees it, and talk to the other agents and the person safely. `tuios --skill <topic>` prints the rest: fleets of agents, turn checkpoints, shipping a worktree, the Inbox and approvals, push notifications, mail, other machines, clients, events, MCP, the tmux shim, herdr's command line, pane grants, configuration, agent features off, errors and recipes.
arch-check
Run the charter architecture fitness check â Checkpoint 7 (Light Verify), alongside `crap-analyzer`. Turns the charter's architectural vision from prose into an objective gate: `dae_arch.py` reads the manifest's `architecture:` rules and reports violations.
auto-arena
Automatically evaluate and compare multiple AI models or agents without pre-existing test data. Generates test queries from a task description, collects responses from all target endpoints, auto-generates evaluation rubrics, runs pairwise comparisons via a judge model, and produces win-rate rankings with reports and charts. Supports checkpoint resume, incremental endpoint addition, and judge model hot-swap. Use when the user asks to compare, benchmark, or rank multiple models or agents on a custom task, or run an arena-style evaluation.
checkpoint
Save session progress by committing changes, pushing to remote, creating or updating a pull request, persisting decisions and patterns to memory layers, compiling session briefing, and archiving completed features. Use when saving work, creating a PR, preserving session state, or manually checkpointing progress.
~aod-build
Generate standardized checkpoint reports for multi-phase implementation projects. Use this skill when pausing implementation at strategic milestones (phase completion, user story completion, critical features) to create comprehensive progress reports with task breakdowns, metrics, knowledge base entries, and resume instructions.
Io.Github.Archetypal Ai/Archetypal Ai
Persistent memory for AI agents. recall, remember, checkpoint — soul preservation.
awrshift
Adaptive decision framework â one dynamic flow with user checkpoints at every phase. Guides from problem to solution through structured research, metrics, factcheck, and sandbox testing. Use when you face a non-trivial decision, need to research before building, plan a feature or experiment, evaluate trade-offs, or the user says 'awrshift', 'let's think this through', 'research first', 'experiment', 'investigate', 'what's the best approach', 'compare options'. Also trigger on: 'иÑÑледÑй', 'ÑазбеÑиÑÑ', 'пÑоанализиÑÑй', 'ÑкÑпеÑименÑ', or when starting any new project phase, migration, launch, or architecture decision. Do NOT use for simple tasks with clear instructions â just do those directly.
a* (autostar)
A generalised autonomous optimisation loop â soft RLVR for the masses. The user defines a goal; the system runs structured experiments, evaluates progress across independent tracks, reflects at strategic checkpoints, and learns from every attempt â including learning how to learn better the next
swarm-checkpoint
Run the project's validation (/check), and on green stage + commit your changes using the project's /commit conventions; on red, report a blocker and note the Queen without committing. Use when the user asks to checkpoint, save progress, or commit safely mid-task.
Checkpointing â Full Session Recording and Pattern Discovery
**Record all session activity and discover reusable patterns. Run everything, every time.**