AI Agent Workflow Collection

Report harness context overhead (KB, words, estimated tokens) for a given path by walking the directory tree and collecting all CLAUDE.md, agents.md, and @-import files the Claude Code harness auto-loads. Use when the user types /context-cost or /ctx-cost, asks how much context overhead a directory has, or wants to know what files the harness is auto-loading.

Map Claude Code harness context overhead across the ENTIRE AIOS as an ASCII tree — brute-forces every directory under $HORIZONROOT, attributes each auto-loaded CLAUDE.md / agents.md / @-import file to the directory it lives in, and reports per-directory, subtree, and grand-total context cost. Use when the user types /context-cost-tree, wants a whole-AIOS context map, or asks where context cost accrues across the tree (as opposed to /context-cost, which measures one path's upward load).

Convert a natural-language prompt into a SAILL agent-team flow — roles with model groups, control-flow flags (parallel/wait/Loop/if needed/if asked/ask user), boxes for sub-teams and nesting, and -context- for run-time values. Use when the user types /convert-prompt-to-saill, says "turn this into SAILL", "convert my prompt to SAILL", or "express this as an agent team / SAILL flow".

Provision a new Horizon AIOS brain — creates the OS user account, the shared brains group and a per-brain group, the workspace folder, a login shell profile, and stores an auto-generated password in the OS keystore. Use when the user types /create-brain, asks to "create a brain", "provision a brain account", "stand up a new brain", or "add a brain user".

Run an iterative documentation/implementation consistency validation pass against the AIOS consistency-check standard, fixing unambiguous drift and surfacing decisions until 100% clean or blocked on user input. Use when asked to run a consistency pass/check, validate AIOS standards, or verify docs and implementation are consistent (e.g. after a feature lands).

Create or rebuild the Horizon AIOS documentation index so every document is registered and referenceable by a stable index entry. Use when asked to index the docs, update/rebuild the documentation index, or after adding/moving/removing documentation (supports CC-G4 in the consistency-check standard).

Run a consistency pass between the Horizon AIOS operational wiki and the underlying source documentation, fixing unambiguous drift and surfacing decisions until clean. Use when the user types /horizonaioswikiupkeep, asks to "update the wiki", "sync the wiki with the docs", or "check if the wiki is still accurate".

Check whether a repo is ready to deploy as a Horizon AIOS Options Package. Points the static readiness checker at an on-disk repo or a git URL and reports, rule by rule (PKG-), whether it meets the Options Package Readiness Standard. Use when the user types /horizon-options-package-readiness, asks "is this ready to be a package", "check my options package", "will this deploy as an AIOS package", or before publishing/installing a new package.

Run the Horizon AIOS index / context-cascade consistency checker — verifies every skill directory is indexed, every documentation file is indexed, every path pointer and $HORIZON variable in the always-loaded context set resolves, and every referenced slash-command exists, then reports passed/warnings/failures. Use when the user types /index-check, asks to "check the indexes", "find dead pointers", "verify the skill indexes", or after adding/renaming/moving a skill, a doc, or an entry in agents.md.

Fetch CURRENT model information from live provider docs (Anthropic, OpenAI, Google Gemini, Ollama) and return a structured catalog the user can use to populate or validate the Horizon AIOS model-preference config. Use when the user types /model-catalog-refresh, or says "refresh model catalog", "update model groups", "check current models", "what models are current", or "validate my model config".

Configure or inspect the in-context model-preference layer — model groups (incl. local/BYO models), per-session slots, and task-class routing — by editing the gitignored extend file. Use when the user types /model-prefs, asks to "set up model groups", "define lowcost", "route docs to a cheap model", "use a local/Ollama model for X", or asks which model a group resolves to.

Audit AIOS skills for their model-preference group callout and assign or refresh it — add a "Model preference" line to skills that lack one, change the group where behavior should differ, and keep the skill indexes' Model-group columns in sync. Use when the user types /model-prefs-assign, says "assign model groups to skills", "audit skills for model preference", "which skills are missing a model group", or after adding/changing skills.

Test which model the model-preference config actually resolves to per group, and (in live mode) spawn small test agents by group name to confirm the spawn honors it. Use when the user types /model-prefs-test, or says "test my model prefs", "check which model lowcost uses", "validate model groups resolve", or "do the model-prefs reliability test".

Create and maintain durable, multi-session objectives that hold long-term context across sessions. Use when the user types /objective, asks to "create an objective", "make a new objective", "list objectives", "show objective N", or "update objective N". A bare /objective (no args) surfaces the conversation's active objective (from a handoff or an earlier mention), else lists existing ones, else offers to create one. Complementary to /handoff — handoffs are point-in-time, objectives are the durable goal they chain back to.

Post-update reconciliation for a deployed Horizon AIOS install. After pulling a sizable AIOS update, diff what the update changed, classify changed files into inert-in-tree vs needs-reapply, print a prioritized reapply checklist, and drive the reapply steps by delegating to /doctor, /harden, /resync-user-skills, and the maintenance-schedule registration script. Invoke after a big update or periodically; it is a manual tool, not a scheduled job.

Validate that the AIOS repo ships every tool needed to run its full lifecycle (install, create brain, create a second AIOS, switch, delete) on each major platform, then emit a ready-to-run admin/sudo test prompt per platform. Use when the user wants a pre-flight tooling check before testing a fresh clone on another machine, or types /pre-flight-tooling-validation.

Deprovision a Horizon AIOS brain — removes its OS user account, per-brain group, workspace folder, profile config, and stored credential, leaving the shared brains group intact. Use when the user types /remove-brain, asks to "remove a brain", "delete a brain account", "deprovision a brain", or "tear down a brain user".

Report the skill inventory and whether the owner's aggregated view (skillsbin + usrskills linked into skillssbin) is in sync with the filesystem and this session, then heal any drift by rebuilding the links. Use when the user types /resync-user-skills, asks what skills they have or whether skills are in sync, adds/removes a skill, or after an upstream sync may have wiped the links.

Evaluate every file in the Terseness Contract Index for context overhead — flags verbose prose, redundant rationale, unnecessary examples, and misplaced content that should be on-demand. Reports FAIL/ADVISORY findings with file:line evidence and concrete cut suggestions. Use when asked to check context overhead, run a terseness pass, or when the consistency check delegates CC-T2.

End-to-end self-test of the Agent Teams system — walk every defined team (or one named team) and spawn each role so it echoes a self-generated nonce, its role, what it was told to do (its charter), and the model it actually ran as. Use when the user types /test-agent-teams --all-teams (all) or /test-agent-teams (one team), says "test the agent teams", "run the agent-team self-test", or wants to verify team resolution + model routing. The skill generates its own nonce.

Browse the Horizon AIOS operational wiki. Bare /userguides spawns one Haiku agent per section to summarize all sections in parallel, then presents the summaries in order. /userguides displays the full text of section N. Use when the user types /userguides, asks for a wiki overview, or wants to read a specific wiki section.