agentleFS
YomudoglyYomudogly/oh-my-agy/skills/hud/SKILL.md

hud

Skill4 starsChanged todayMITView on GitHub

[oh-my-agy] Show an on-demand status board of active oh-my-agy workflows (autopilot phase, ultragoal progress, team tasks, QA cycle) from .omagy/ state, or create a structured, self-contained session handoff artifact under .omagy/artifacts/handoff-<slug>-<UTC>.md (--handoff) so work resumes cleanly after /compact or in a fresh conversation. Use for $hud, 'what modes are active', 'workflow status', 'create a handoff', 'save context for next session', or before context compaction.

10 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use visual-verdict, autoresearch and 1 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • LucasCooperBeySkill

    visual-verdict

    Structured visual QA verdict for screenshot-to-reference comparisons. Returns JSON scores driving edit iterations. Use after any UI implementation to verify visual fidelity.

    Also in 3 projects that use this

  • GitHubSkill

    autoresearch

    Autonomous iterative experimentation loop for any programming task. Guides the user through defining goals, measurable metrics, and scope constraints, then runs an autonomous loop of code changes, testing, measuring, and keeping/discarding results. Inspired by Karpathy''s autoresearch. USE FOR: autonomous improvement, iterative optimization, experiment loop, auto research, performance tuning, automated experimentation, hill climbing, try things automatically, optimize code, run experiments, autonomous coding loop. DO NOT USE FOR: one-shot tasks, simple bug fixes, code review, or tasks without a measurable metric.

    Also in 3 projects that use this

  • GoogleSkill

    tdd

    Test-driven development. Use when the user wants to build features or fix bugs test-first, mentions "red-green-refactor", or wants integration tests.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

1 rule. 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 0

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If nothing is active, print [OMA] No active modes and suggest $autopilot or $ultragoal.

    line 62

“Common” is measured: kept by at least 6 in 10 of the 7 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

HUD (Status Board)

In Antigravity, oh-my-agy provides both an on-demand chat status board and a live Auxiliary Pane UI extension (sidecars/hud/sidecar.json) accessible via the sidecar:// link scheme.

Aux Pane UI Pill (sidecar://oh-my-agy/hud/)

When sidecars/hud/sidecar.json is present and the plugin is enabled, include a single one-click pill at the end of the HUD output (following Antigravity's ui-plugin-navigation skill rules: one single SIDECAR_UI_ENTRYPOINT_AUX_PANE view in sidecar.json so the Extensions menu shows only Oh My Agy, with internal tabs inside the panel):

  • [oh-my-agy HUD](sidecar://oh-my-agy/hud/) (served by sidecars/hud/main.mjs + sidecars/hud/ui.mjs + sidecars/hud/client.mjs with live GET /api/events SSE updates, 15 s fallback polling, digest-diffed stateful DOM preservation, Cmd/Ctrl+K Command Palette #cmd-palette, top-bar workspace ?activeRoot= selector & Handoff button, wide Slide-Over #artifact-drawer with asymmetric transitions & code-block copy, GPU sliding tab pills, arrow-key tablist navigation, keyboard shortcuts Cmd/Ctrl+K / / / 1–5 / Escape, and light/dark CSS tokens across 5 tabs: Overview with workspace/scope filter chips, single Current Workflow Mode spotlight & contextual remediation buttons, GPU scaleX progress bars, Team Task DAG waves & JSON Receipt previews, 5-event Recent Ledger Events, 2-tab Reviews/Research switcher, 10-item collapsible Plans, Specs, Handoffs & Artifacts, quick actions, and Benchmarks/Visual-Ralph/Loft/Code-Health/Stitch DESIGN.md/Forensics/Notifications/Doctor cards; Workflows horizontally scrollable animated step pipelines; searchable Skills catalog with <mark> match highlights, --flag composer chips & Preview drawer for SKILL.md + references/*.md; Subagents roster with Model tier pills inherit|flash_lite|flash|pro & invoke_subagent snippet preview; and interactive 3D Canvas mind-map / List [[slug]] Markdown Wiki explorer with /api/wiki-search, TOC jump chips, computed backlinks, omagy_wiki/log.md activity, and left-rule Working Agreements)

Desktop-Only HUD Surfaces (Antigravity, Claude Desktop & Codex Desktop)

The HUD workbench runs exclusively on Desktop apps (no CLI statusLine/tmux, IDE extensions, or MCP Apps ui://):

  • Antigravity Desktop: [oh-my-agy HUD](sidecar://oh-my-agy/hud/) via sidecars/hud/sidecar.json. Defaults host to 'antigravity'.
  • Claude Desktop: Configured via sh hooks/run.sh host-setup claude in <workspace>/.claude/launch.json ("oh-my-agy-hud" on loopback port 48800..49799) or auto-spawned on SessionStart via sidecars/hud/main.mjs desktop-serve --host claude on 47800..48799. Interactive actions copy prompts cleanly (clipboard-handoff), and subagent-standalone executes claude -p --agent <role>.
  • Codex Desktop: Auto-spawned idempotently on Desktop SessionStart via sidecars/hud/main.mjs desktop-serve --host codex on loopback port 47800..48799 for Codex Desktop's in-app Web Preview. POST /api/dispatch supports loopback RFC 6455 Codex app-server JSON-RPC (OMAGY_CODEX_APP_SERVER_URL), codex exec (subagent-standalone), and clipboard-handoff.
  • Desktop Gating & Security Controls:
    • OMAGY_DESKTOP_HUD=1 forces auto-start on 'claude'/'codex' (suppressed on plain CLI, VS Code, Cursor, JetBrains, and CI); OMAGY_DESKTOP_HUD=0 disables Desktop auto-start completely.
    • OMAGY_HUD_IDLE_TIMEOUT_MS configures idle shutdown (default 2 hours; postponed while /api/events SSE clients remain connected).
    • desktop-serve validates loopback Host headers (127.0.0.1, localhost, [::1], 403 otherwise) before injecting dynamic <meta name="omagy-hud-token"> on GET /, stores its 0o600 registry under runtimeDir(), and omits ?token= from SessionStart output.

Presets

  • minimal: one line, e.g. [OMA] autopilot:ultragoal | ultragoal:3/7
  • focused (default): the minimal line plus each active mode's phase, iteration, and last-update age.
  • full: focused plus goal/task tables and recent ledger entries.

Accept $hud --preset=<name>. A default can be stored in .omagy/config.json as {"hud": {"preset": "focused"}}.

Sources (strictly read-only)

  • Zero-mutation status invariant & live SSE (/api/status & /api/events): $hud, GET /api/status, and debounced GET /api/events (150 ms .omagy/ + omagy_wiki/ watcher debounce with computeSnapshotDigest duplicate suppression and statusRequiresToken verification) are strictly observational — never write state files or mutate .omagy/ while rendering. Interactive actions (Run in Chat, New Chat, Dispatch into Active Chat, Launch Standalone Chat, Continue, two-step Cancel Mode, Create Handoff, Run /doctor) use POST /api/dispatch via Antigravity's native agentapi CLI without mutating .omagy/ files directly, while window.addEventListener('message', ...) rejects null/foreign event.source frames and reconnects /api/events on workspace-change.
  • .omagy/state/*-state.json and .omagy/state/sessions/*/*-state.json: mode, active, current_phase, iteration, updated_at (aggregate across the 5 newest session directories by default, or filter to a requested --session=<id>).
  • .omagy/ultragoal/goals.json and the tail of ledger.jsonl: goal counts by status, current goal
  • .omagy/state/team/*/tasks.json: task counts by status, worker assignments. Show active team boards even when no session team-state.json snapshot exists, and hide completed team runs whose updated_at is older than 20 minutes in team-only views.
  • .omagy/goals/performance/*/state.json: baseline vs lastValidation
  • Code Health & Technical Debt (analyzeDiffCodeHealth, scanShortcutDebt & codeHealth.floorGuard): working-tree function reuse ratio (used vs new symbols, reusePct), open callers of removed symbols (⚠ N open callers), the repository shortcut: / ceiling: debt ledger (total markers and no-trigger rot risk count), and 5,000ms + .git/index & .git/HEAD mtime-cached codeHealth.floorGuard anti-tampering telemetry (runFloorGuard: ok, exitCode, deletedTests, removedAssertions, untrackedBlockers, suppressions, emptyCatches, protectedBlocksModified, total).
  • Diagnostics & Runtime Health (validateQualityGates & doctor.skillEval): configured .omagy/config.json quality gates, stale mode detection, recent .omagy/logs/hook-errors.jsonl fail-open entries, and mtime-memoized doctor.skillEval routing health (evaluateSkillRouting: ok, rank1Pct, minRank1, collisionErrors, negativeSurvivalPct) surfaced as Skill Eval: PASS (<rank1Pct>%) | FAIL.
  • Safe Read-Only Artifacts, Skill Docs & Wiki Search (/api/artifact, /api/skill-doc, /api/wiki-search + #artifact-drawer): resolveSafeArtifactFile serves symlink-free allowlisted files under .omagy/{artifacts,plans,specs,evidence,ultragoal,goals} (including JSON Receipt previews .omagy/evidence/*.json and .omagy/plans/ralplan-review-*.json, .omagy/artifacts/{visual-ralph,loft,research,ideas}, top-level code-review-*.md / handoff-*.md / ask-*.md, .omagy/plans/*.md, .omagy/specs/**/*.md, and Loft (?:variant-[a-c]/)?index.html + local .js/.mjs/.css assets) with X-Content-Type-Options: nosniff, Content-Security-Policy: sandbox, and redactSecrets() applied to .md/.txt/.json responses. resolveSafeSkillDoc serves GET /api/skill-doc (symlink/realpathSync-contained SKILL.md + references/*.md playbooks from skills/<dir>/ for the Skills Preview button & drawer chips), and GET /api/wiki-search runs token-gated queryWikiPages. #artifact-drawer renders JSON receipts, a 3-mode Visual Ralph comparator (side-by-side, split-slider, onion-skin), an opaque-origin srcdoc Loft <iframe sandbox="allow-scripts" referrerpolicy="no-referrer"> (never allow-same-origin, zero token in iframe URL, external scripts/links stripped, local assets & relative ES module imports bundled, active HUD theme variables injected), and XSS-escaped Markdown previews with SDO section headings.
  • Workbench Governance Cards (Overview): Surfacing mtime-cached Stitch DESIGN.md WCAG contrast audits (validateDesignMd), Session Transcript Forensics (scanSessionTranscripts, with default-collapsed Top Failure Clusters), and boolean-only Notification readiness ({ slack, discord, telegram, webhook, command }).
  • manage_subagents list and manage_task list: live subagents and background tasks

Rendering

[OMA] autopilot:ultragoal (iter 1, 4m ago) | ultragoal:3/7 complete | team:qa-sweep 2/5 | tasks:1 bg

Then, for focused/full, a small table:

ModePhaseIterUpdatedNotes

Flag problems:

  • stale: active and updated_at older than 6h. The Stop hook ignores it; suggest $cancel or resuming.
  • stuck: two recent consecutive continuation nudges without a state change. Check /tmp/oh-my-agy/*.nudge.json, or $TMPDIR/oh-my-agy/ on macOS.
  • inconsistent: e.g. autopilot at ultragoal with no goals.json.

If nothing is active, print [OMA] No active modes and suggest $autopilot or $ultragoal.

For a live view, the user can ask you to schedule a recurring refresh (e.g. every 5 minutes while a long run is active).

Structured Session Handoff ($hud --handoff)

While default $hud status rendering is strictly observational, explicit $hud --handoff writes a self-contained continuation artifact before /compact or switching to a fresh conversation:

  1. Inspect active state & git status (redacting any secrets via redactSecrets or running sh <plugin>/hooks/run.sh handoff when available):
    • Active state files in .omagy/state/ and .omagy/state/sessions/<conversationId>/.
    • .omagy/ultragoal/goals.json and the tail of .omagy/ultragoal/ledger.jsonl if present.
    • git status --short and git diff --stat to capture modified and untracked files.
  2. Write the handoff artifact:
    • Derive <slug> and UTC timestamp (YYYYMMDDTHHMMSSZ), write .omagy/artifacts/handoff-<slug>-<UTC>.md, and update .omagy/artifacts/handoff-latest.md:
# Session Handoff: <title>

- **Created**: `<ISO UTC>`
- **Conversation ID**: `<conversationId>`
- **Active Mode / Phase**: `<mode> (<current_phase>) | none`
- **Branch / Commit**: `<branch> @ <short-sha>`

## 1. Objective & Current Status
<1-3 sentences stating the target outcome and what percentage/stage is complete>

## 2. Completed Work & Verification Evidence
- `<file:line>` — <change summary> (Verified by: `<command>` -> exit `<code / output>`)

## 3. Active Durable State & Artifacts
- State file(s): `.omagy/state/...`
- Plan / Spec: `.omagy/plans/...` or `.omagy/specs/...`
- Ultragoal ledger: `.omagy/ultragoal/goals.json` (`<completed>/<total>` goals complete)
- Working agreements / Wiki updates: `omagy_wiki/...`

## 4. Key Decisions & Adopted Defaults
- `[from-user]` / `[adopted-default]`: <decision + 1-line rationale>

## 5. Open Blockers, Dirty Files & Residual Risks
- Uncommitted files: `<list from git status>`
- Open blockers / next discriminating probe: `<none or exact details>`

## 6. Exact Resume Prompt (Copy-Paste into Next Session)
> Read `.omagy/artifacts/handoff-latest.md` and resume from `<exact next step or $skill command>`.
  1. Report: output the clickable link to .omagy/artifacts/handoff-<slug>-<UTC>.md, display the Exact Resume Prompt, and remind the user they can now run /compact or open a new session safely.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.