tackroom
tackroom is a public Go CLI. The user's canonical configuration is a separate private directory or repository.
Config precedence:
--config /path/to/tackroom.yaml$TACKROOM_HOME/tackroom.yaml~/.agents/tackroom.yaml
Never infer configuration from the current project. Never replace a user's config root with a public checkout.
Commands
tackroom setup [--from GIT_URL] [--overlay NAME] [--review] [--verbose] [--memory off|basic|memsearch] [--agents ...] [--no-cron] [--dry-run] [--json]
tackroom status [--verbose] [--exit-code] [--agents ...]
tackroom sync [--pull] [--dry-run] [--replace-conflicts] [--agents ...] [--allow-fresh[=item,...]] [--reason ...]
tackroom sync --plan-digest DIGEST # sync only if the plan and the config are still the ones sync --dry-run --json showed
tackroom doctor [--e2e] [--agents ...]
tackroom doctor audit [--json] # findings per item; exit 1 on an unaccepted CRITICAL in external content
tackroom doctor audit accept <item> <rule> --reason "..." # record an audit decision; item is kind:name, e.g. skill:foo
tackroom doctor audit revoke <item> <rule>
tackroom config validate
tackroom config print
tackroom config inventory # the web view's inventory as JSON: harnesses, skills, roles, MCP servers, hooks
tackroom config set <section/name/field> <value> # one existing field, e.g. mcp_servers/exa/enabled false; sync applies it
tackroom view [--addr 127.0.0.1:8765] [--no-open] [--secure-cookie] [--ssh-host user@host] [--token-file PATH]
tackroom tui [--config PATH] # deprecated, removed in 3.0; do not recommend it: use view, status or cron collect --table
tackroom skill new <name> [--description ...]
tackroom skill list [--agents ...]
tackroom skill info <name>
tackroom skill sources [--json]
tackroom skill sources search <query> [--json]
tackroom skill sources add <owner/repo[/path] | git URL> [--skills a,b | --all] [--latest] [--name N] [--no-sync]
tackroom skill update [name ...] [--allow-fresh[=item,...]]
tackroom publish [--target NAME] [--skills a,b] [--dry-run] [--json] [--yes]
tackroom mcp <list|add|import|remove> [options]
tackroom restore [<backup> | latest] [--dry-run]
tackroom tern [--config PATH] [--dry-run] # open a tackroom workspace in the Tern terminal
tackroom rex [--config PATH] [--dry-run] # open a tackroom workspace in the Rex terminal
tackroom pull [options] # git pull --ff-only, then sync; what the periodic job runs
tackroom cron [--interval 15m|--deps|--remove] # periodic sync job, installed by setup; default 15m
tackroom hosts [HOST...] [--json] [--config PATH] # tackroom on each machine in crons.hosts: version, sync state, config commit, fix
tackroom hosts fix [HOST...] [--dry-run] # run the fix hosts names (setup, pull or sync there); never commits or pushes
tackroom hosts setup HOST [--from GIT_URL] [--overlay NAME] [--no-cron] [--install] [--dry-run] # tackroom setup --from on another machine over SSH
tackroom cron collect [--hosts local,HOST...] [--table] # read-only scheduled-job inventory
tackroom cron apply [--dry-run] # install crons.jobs entries with schedule+command as launchd, systemd or crontab units
tackroom cron run|pause|resume [--host HOST] ID # act on a launchd job, here or over SSH
tackroom cron schedule ID "MIN HOUR DOM MON DOW" # new schedule for an owned job on this machine
tackroom cron add ID --schedule "MIN HOUR DOM MON DOW" --command CMD [--arg ARG ...] [--host NAME] [--effect TEXT] [--local] [--dry-run] [--no-apply] # declare an owned job, then apply
tackroom cron edit ID [--schedule EXPR] [--command CMD [--arg ARG ...]] [--host NAME] [--effect TEXT] [--dry-run] [--no-apply] # change fields of an owned job
tackroom cron remove ID [--dry-run] [--no-apply] # delete an owned job's declaration and unit
sync prints the skills, agents, MCP servers, hooks, packages and plugins the run changed (harnesses with identical changes grouped), a table with one row per harness, and every managed item per harness, with long lists wrapped to the terminal. status stays compact; --verbose adds the full lists. Both drop escape codes when output is not a terminal and color when NO_COLOR is set.
view (browser web UI) is the primary surface, on a desktop, a phone or a narrow
pane: a dashboard with per-agent sync state, outside sources with their pins, holds
and blocks, skill and role matrices, MCP and hook targeting per agent, unmanaged
items in agent folders, and a YAML editor. The CLI is the engine: every view action
calls the same code, and the view has no data or action the CLI lacks. It edits shared YAML or the
machine-local overlay; effective configuration is read-only. Each toggle
applies immediately, and view never runs sync implicitly. It binds only to
loopback and uses a session cookie plus CSRF and origin protection. config
validates or prints the result. AgentsView (agentsview serve) is an optional
companion tool that you run on its own.
Run tackroom help --all for maintenance commands.
setup
tackroom setup asks nothing and never reads stdin. It:
- Creates the canonical config root when absent.
- Extracts the public starter assets into a config root that has no
tackroom.yamlyet. An existing config root gets no starter files, so a starter role, skill orplugin.jsonthe user deleted stays deleted; only a memory tier that is on gets its missing files undermemory/. - Detects supported harness binaries.
- Copies every native skill, role, and MCP server into the config root. Where harnesses have different copies of a skill, the first detected harness's copy is kept. A role is shared only when every harness's copy would come back unchanged from the shared one (same fields and values, same instructions, nothing added); a role with a field tackroom cannot write, without frontmatter, with differing copies, or that is a symlink stays each harness's own, and no starter role of that name is added. Imports are copy or conversion only and add no fields.
- Registers the memory tier (
basicon a new config; an existing config keeps its tier unless--memoryis given).tackroom.yamlis rewritten only when its content changes. - Patches only the required native harness settings.
- Runs the first sync: every native file or skill folder it replaces, identical or not, is backed up to
~/.local/state/tackroom/backups/<time>/first, sotackroom restore latestundoes it; a native role is replaced only when the rewritten role is the same role, never a symlinked one, and a role conflict leaves only that file alone while the rest of the harness syncs. Removals or overwrites of content an earlier tackroom run wrote are left fortackroom sync. Instructions merged from several agents are listed as needing attention. - Initializes and commits git on a first setup (no
tackroom.yamlbefore, not already under git). - Installs the periodic sync job (
tackroom pullevery 15 minutes), unless--no-cron. - Prints a short summary ending in
Everything is synced.or a list of what needs the user, each with its fix command, and then exits 1 (the sync, commit and job may already be done). A rerun on a set-up machine printsAlready set up; everything is synced.or what changed.
--review brings back the interactive review screen (share, keep harness-specific, or skip per item) and a prompt before git init, conflict replacement and destructive changes; without a terminal it falls back to sequential [y/N] prompts. --verbose prints every step and the full sync report. --dry-run prints what the run would import, whose copy it keeps and which roles stay each harness's own, and changes nothing; --json emits the detection result and exits. --yes is accepted as a no-op.
On another machine, tackroom setup --from <git url> clones the user's config repository into the config root when it is missing or empty, or pulls it with git pull --ff-only when it is already a clone of that URL, then sets up from its tackroom.yaml without adding starter files. A root with other content, another origin, or no origin is refused unchanged. Without --memory, setup never changes an existing config's memory tier; a memsearch config on a machine without memsearch keeps its hooks and the summary lists the install command.
For a machine of a known kind (a cloud VM, say), tackroom setup --from <git url> --overlay cloud --no-cron also copies the committed overlays/cloud.yaml, a complete tackroom.local.yaml, into place before the first sync. An unknown name stops setup before it syncs anything and lists the overlays the repository has. An existing tackroom.local.yaml that differs is never overwritten: setup syncs with it, lists it as needing the user and exits 1. periodic_sync is not compared, so a rerun after --no-cron reports nothing to do. Only setup --overlay reads overlays/.
Setup also imports existing instructions files (~/.claude/CLAUDE.md, ~/.codex/AGENTS.md, ...) into ~/.agents/AGENTS.md, then backs up the originals to ~/.local/state/tackroom/backups/<time>/ and links them; several different ones are merged under a ## From <agent> (<file>) heading each, and every setup run lists them as needing attention until the user merges them and deletes those headings.
Extra always-on files (voice, user profile) go under instructions: in tackroom.yaml, never pasted into AGENTS.md. Every managed root doc then carries AGENTS.md plus those files: Claude Code via @ imports, the others as a generated copy marked Generated by tackroom. Edit the sources, then run tackroom sync; status shows a stale copy as drifted.
Running setup for a user (agents)
tackroom setup --dry-run # show what would be imported; changes nothing
tackroom setup # imports, git init, backs up and replaces differing files; no prompts
tackroom status --json # verify; exit 0 = everything synced
Exit codes: status and sync exit 1 when anything is out of sync or conflicting; doctor exits 1 when any check fails or warns. All exit 0 otherwise. Warnings are advisory (for example a package younger than three days, or a registry that could not be reached); to gate on real failures, read failed from doctor --json. --json output on those three is a stable contract: synced, agents[].conflicts, agents[].skills|mcp|roles|hooks.{managed,missing,drifted}, plus agents[].mcp.stale and agents[].hooks.stale when present. When sync reports conflicts, show the listed paths to the user before running tackroom sync --replace-conflicts.
The public starter contains tackroom, six generic roles (architect builder general researcher reviewer tester), and reusable memory scripts. Personal skills, hooks, MCP servers, secrets, and memory data belong only in the private config repository.
Memory tiers:
off: no managed memory hooks.basic(default): Python-3-only bounded Markdown digests under$KNOWLEDGE_DIR/sessions/plus recent context at session start.memsearch: the indexed SessionStart, Stop, and SessionEnd pipeline; requiresmemsearchonPATH.
Verify the complete setup with tackroom doctor --e2e.
status
Reports each configured harness as detected or not detected and compares the four managed surfaces with native state:
- skills
- MCP servers
- hooks
- agent roles
Missing, drifted, conflicting, stale managed, and unrelated external entries are reported separately. --exit-code exits 0 in sync, 1 on drift, 2 on error; held and blocked items do not count as drift.
sync
Reconciles only configured managed entries. Unrelated native content remains untouched.
For symlink-based harnesses, skills point to canonical directories under ~/.agents/skills. Qwen Code uses skills.directories: ["~/.agents/skills"] and consumes the canonical tree without creating a duplicate mirror. OpenCode, Cursor, Gemini CLI, GitHub Copilot CLI, Grok Build and DeepSeek Harness read ~/.agents/skills themselves; tackroom mirrors into their skill roots only when the config root is elsewhere.
Agent roles are canonical Markdown files under ~/.agents/agents/ and render to:
- Claude Code:
~/.claude/agents/<name>.md - Codex:
~/.codex/agents/<name>.toml - Factory Droid:
~/.factory/droids/<name>.md - OpenCode:
~/.config/opencode/agents/<name>.md - Pi with
pi-subagents:~/.pi/agent/agents/<name>.md - OMP:
~/.omp/agent/agents/<name>.md - Qwen Code:
~/.qwen/agents/<name>.md - Cursor:
~/.cursor/agents/<name>.md - GitHub Copilot CLI:
~/.copilot/agents/<name>.agent.md - Grok Build:
~/.grok/agents/<name>.md - Gemini CLI:
~/.gemini/agents/<name>.md(local subagents)
Pi always has managed skills. With pi-subagents installed, tackroom renders canonical roles into Pi's user agent directory. It patches canonical and Agent Plugin MCP entries into ~/.pi/agent/mcp.json, which Pi 1.0 and later read natively (older Pi needs pi-mcp-adapter). Put pinned package sources under the Pi target's packages list to make sync reconcile ~/.pi/agent/settings.json; Pi installs missing declared packages at startup. Tackroom does not install the Pi executable. The role files remain inert without pi-subagents. OMP is a separate target.
For MCP servers, sync patches only named canonical entries and preserves unrelated native servers. A server sync wrote earlier and that is no longer wanted for an agent (removed from tackroom.yaml, disabled, or dropped from its agents) shows as mcp stale in status and is removed by the next sync, after the native config file is copied to ~/.local/state/tackroom/backups/<time>/. Sync never claims an entry the agent already had; mcp remove also claims native entries that still match the definition it removes. Import redacts literal environment values to ${KEY} references and names the redacted keys; the source agent keeps its values. Other target agents get the literal ${KEY}, which they may not expand: put the real value in their native config (sync never replaces a value with a reference) or import with --agents <source>. List output never prints values.
For hooks, sync registers only declared entries on harnesses with verified hook support. A declared hook that is turned off, or whose agents no longer names an agent, shows as hooks stale in status while that agent still has it registered and is removed from that agent by the next sync; registrations of hooks the config does not declare are left alone. Host-local review and approval state remains outside tackroom.
Claude Code plugins (mods, workflows, monitors, output styles, LSP, bin/) stay in Claude's native layout under the config root, listed in <config-root>/.claude-plugin/marketplace.json. sync registers that directory as a marketplace in ~/.claude/settings.json (extraKnownMarketplaces) and enables each relative-path entry (enabledPlugins). Claude Code reads it in place: edit, then /reload-plugins. User disables (kept across a marketplace rename) and defaultEnabled: false are respected; stale enable keys are pruned and removing the marketplace file removes the registration. A name already registered with another source, or a malformed file, is a conflict and nothing is written. doctor runs claude plugin validate on it when Claude Code is detected.
Codex gets an agent-plugins-spec external source as a native plugin when it sets mcp: true, has a valid root plugin.json, targets Codex (mcp_agents empty or listing it, exclude_agents not listing it), keeps the default skills/ layout (no skills:, skill_dir: or skill_dirs:) and is not held or blocked. sync writes a generated marketplace tackroom at ~/.agents/external/.agents/plugins/marketplace.json, registers it with codex plugin marketplace add, installs with codex plugin add <name>@tackroom when the plugin is missing or Codex's copy differs, and removes *@tackroom plugins that are no longer declared with codex plugin remove. Codex then loads that source's skills and MCP servers from the plugin alone, so tackroom links no skills and writes no [mcp_servers] rows for it. enabled = false under [plugins."<name>@tackroom"] in ~/.codex/config.toml is respected and shown as plugins off. status, sync --dry-run and doctor never run codex. Tested with codex-cli 0.159.2.
sync --pull runs git pull --ff-only in the private canonical repository before reconciliation.
sync --dry-run prints the plan a real sync would apply, including held and blocked items, and writes nothing (it fetches nothing and makes no registry lookups). sync --allow-fresh releases held fresh items, all or a comma list with --allow-fresh=item,..., optionally recording --reason "..."; held and blocked items do not count as drift.
External skills
Declare external Git sources in the private tackroom.yaml:
external_skills:
- url: https://github.com/example/shared-skills
branch: main
skill_dirs: [engineering/alpha, productivity/beta]
materialize: true
tackroom.lock records exact commits and materialized ownership. sync repairs drift to the pin; skill update explicitly advances it. doctor audits external source content and fails on materialization drift.
Add a source without editing YAML: tackroom skill sources add mattpocock/skills --skills tdd,grill-me (or --all; without either, an attended run lists the skills to pick and an unattended one fails with the list). It pins the newest commit at least 3 days old (--latest pins the head), writes the entry with materialize: true, refuses to overwrite a different ~/.agents/skills/<name> (an identical copy, as npx skills add leaves it, is adopted), then syncs. tackroom skill sources lists configured sources and the ones installed outside tackroom (Claude Code plugins, npx skills); search queries skills.sh. A source that Claude Code already has as a plugin gets exclude_agents: [claude], so Claude does not load its skills twice. Two repos with the same basename need a distinct name:; add picks <owner>-<repo> itself.
publish
publish is the outward analogue of external skills: it pushes canonical skills to a remote skill registry (OpenAI /v1/skills) and pins the returned id/version in tackroom.lock under published_skills. Declare opt-in targets with an explicit skill allowlist in tackroom.yaml:
publish_targets:
- name: openai
kind: openai-skills
enabled: true
skills: [jobs, tech-search]
api_key_env: OPENAI_API_KEY # key is read from this env var, never inlined
tackroom publish uploads only allowlisted skills whose bundle content changed since the last run (content-hash idempotency: create, skip if unchanged, new version if changed). It is inert until a target sets enabled: true. Use --dry-run to preview, --target/--skills to narrow, --json for machine output, --yes/-y to skip the confirmation prompt. A real upload always prints a US-only / no-Zero-Data-Retention warning first; do not publish skills carrying secrets or private vault content.
MCP management
tackroom mcp list
tackroom mcp add local --command uvx --arg pkg@1.2.3 --env KEY=value
tackroom mcp add docs --url https://mcp.example.com/mcp --header 'Authorization: Bearer ${DOCS_TOKEN}' # remote: Claude Code, Codex, Pi, Grok, Cursor, Gemini CLI
tackroom mcp import claude local --agents=codex,droid,pi,omp
tackroom sync
tackroom mcp remove local
tackroom sync # removes it from agents where tackroom wrote it or it still matches
Use versions verified against the package registry. Keep secrets in environment variables or host-local native config, never canonical YAML.
doctor --e2e
Runs sync, status, and doctor as one health check. It fails on drift, conflicts, invalid external pins, unsupported managed claims, or other doctor errors.
tackroom doctor --e2e
doctor audit lists findings per item (skill:, plugin:, role:, hook:, mcp:, package:) and exits 1 when an external item has an unaccepted CRITICAL finding: a prompt-injection phrase, a secret sent out or a credential file read, invisible tag characters, or a download piped into a shell (rule remote-exec). A match inside a Markdown code fence or a script comment counts one level lower. Use doctor audit accept <item> <rule> --reason "..." to record a decision (stored in tackroom.lock under accepted:) and doctor audit revoke <item> <rule> to undo one.
view
Opens the canonical config UI in your browser: the review-first authoring surface, served over a loopback-only HTTP listener embedded in the tackroom binary (no Node or separate daemon). Views: Overview (agent cards with sync state, counts and estimated skill-listing tokens, every outside source as pinned, held or blocked with the command that releases or accepts it, and, with more than one host in crons.hosts, a Machines block with each machine's health, summary and fix, the same as tackroom hosts), Skills and Roles (matrices of what is on disk per agent), MCP & hooks (toggle an entry or the agents it targets), Unmanaged (foreign skills, conflicts, stale links with the next command), Crons (scheduled jobs and failing services on the hosts in crons.hosts, the same data as tackroom cron collect, with Run now, Pause and Resume for launchd jobs, a schedule field for owned ones, and Add cron, Edit and Delete for jobs tackroom owns), and Config (edit, check and save the YAML). It authors the shared YAML and the machine-local overlay, shows a read-only effective merge, guards saves by revision, and keeps sync as a separate preview/confirm step in which every destructive item must be ticked. Loopback-only bind, tokenized startup URL bootstrapped into an HttpOnly, SameSite=Strict session cookie, plus CSRF and origin checks on mutations.
It prints the tokenized URL on its own line and, when running locally, opens it in your default browser. --no-open suppresses the browser launch. --addr sets the loopback bind (default 127.0.0.1:8765). --secure-cookie serves an HTTPS proxy such as tailscale serve: requests through it get Secure cookies and https origin checks, while direct loopback requests keep working over http. The token is kept in view.token under $XDG_STATE_HOME/tackroom (~/.local/state/tackroom) or --token-file, so the URL survives restarts; tackroom view opens an already running view instead of starting another (it finds one at the last recorded address unless --addr is given), and a long-running view restarts itself when the tackroom binary is upgraded. On a remote host, pass --ssh-host user@host (or run inside an SSH session, where it derives the host from SSH_CONNECTION) to print a ready ssh -L tunnel command instead of auto-opening.
tackroom view # open the config UI locally
tackroom view --no-open --addr 127.0.0.1:8765 # print the URL, do not open a browser
tackroom view --ssh-host me@box # remote: print an ssh -L tunnel command
Capability matrix
| Harness | Skills | Roles | MCP | Hooks |
|---|---|---|---|---|
| Claude Code | yes | yes | yes | yes |
| Codex | yes | yes | yes | yes |
| Factory Droid | yes | yes | yes | yes |
| OpenCode | yes | yes | yes | no |
| Pi | yes | yes, via pi-subagents |
yes, native since Pi 1.0 | no |
| OMP | yes | yes | yes | no |
| Qwen Code | yes, config-driven | yes | yes | yes |
| Cursor | yes | yes | yes, ~/.cursor/mcp.json |
yes, ~/.cursor/hooks.json |
| GitHub Copilot CLI | yes | yes | yes, ~/.copilot/mcp-config.json |
yes, ~/.copilot/hooks/tackroom.json |
| Kimi Code | yes | yes | yes, ~/.kimi-code/mcp.json |
yes, ~/.kimi-code/config.toml |
| Grok Build | yes | yes | yes, ~/.grok/config.toml |
yes, ~/.grok/hooks/tackroom.json |
| Gemini CLI | yes | yes | yes, ~/.gemini/settings.json |
yes, ~/.gemini/settings.json |
| Antigravity CLI | yes | no | yes, ~/.gemini/config/mcp_config.json |
yes, ~/.gemini/config/hooks.json |
| DeepSeek Harness | yes | no | yes, tackroom-mcp-* rows in ~/.dsh/cordis.patch.yml |
no |
| Amp | yes, config-driven | no | yes, amp.mcpServers in Amp settings |
no |
Do not add a surface without a verified native adapter and focused tests.