verification-cockpit
Skill2 starsChanged 40 days agoMITView on GitHub
Generates project-specific tmux verification cockpits for repos that use an av-style agent verify workflow. Reads target project AGENTS.md, setup docs, and scripts; writes .agents/verification/ with SOC-style panes and tiered auto-launch. Use when setting up av layouts, mission-control verify panes, or regenerating .agents/verification after stack changes.
In this repository49 other files it gives its agents.
- AGENTS.mdAGENTS.md
- adversarial-auditSkill
- agentic-reactorSkill
- agent-orchestratorSkill
- ai-optimizationSkill
- architecture-synthesisSkill
- audit-allow-buildsSkill
- audit-ide-dependenciesSkill
- author-workflow-skillSkill
- bdd-strategizerSkill
- chrome-extension-mv3Skill
- concurrent-cli-agentsSkill
- control-feederSkill
- control-graphSkill
- cursor-transcript-harvestSkill
- cv-promote-guardSkill
- devcontainer-hardenedSkill
- dual-copy-skill-publishSkill
- explore-repo-readonlySkill
- finder-reactorSkill
- fix-dependency-securitySkill
- fusion-sageSkill
- git-worktreesSkill
- grok-host-prepSkill
- gt-flowSkill
- higher-order-decision-architectSkill
- looperSkill
- master-plannerSkill
- mvu-refactor-planSkill
- odysseus-navigatorSkill
- project-editor-profileSkill
- react-client-expertSkill
- semantic-markup-cssSkill
- semantic-nameSkill
- session-unit-orderSkill
- shell-kernel-ontologySkill
- skill-rename-propagationSkill
- split-to-prsSkill
- src-tree-reorganizeSkill
- stellar-roadmapSkill
- stellar-spacemapSkill
- subagent-delegationSkill
- subagent-explore-reportSkill
- supply-chain-hardenSkill
- tauri-agenticSkill
- tauri-ipc-debugSkill
- tidy-commit-pushSkill
- upgrade-packagesSkill
- x-agent-resourcesSkill
The file
We split this skill into the rules it gives an agent
8 rules. About 0 are common, 4 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 54
Read in order (stop when you have enough signal):
- Very specific: line 74
Rule: if it writes deps, data, or project structure → mutate.
- Distinctive: line 80
Mandatory: every pane answers *what failure does this surface?* If it does not, omit it.
- Very specific: line 82
Golden ratio: all splits use φ ≈ 1.618 → 62% major / 38% minor (bin/lib/verify-layout.sh).
- Very specific: line 96
Use verify_layout_build_golden_grid from verify-layout.sh for the default 4-pane skeleton.
- Very specific: line 151
Use templates from templates/ in this skill directory.
- Distinctive: line 156
Must run inside tmux
- Distinctive: line 173
If no cockpit section exists, add under setup/verify:
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Verification Cockpit Generator
Portable skill — install from the agent skills library (verification-cockpit/). Primary use: generate layouts in app/library repos you are verifying.
Output: .agents/verification/ in the workspace you are verifying. av auto-delegates when tmux-layout.sh exists there.
Adopt in a target project
Copy or symlink this skill into the project you are verifying:
SKILLS_ROOT=~/skills # or wherever you cloned the skills library
mkdir -p /path/to/my-app/.cursor/skills
ln -sfn "$SKILLS_ROOT/verification-cockpit" /path/to/my-app/.cursor/skills/verification-cockpit
# or: cp -a "$SKILLS_ROOT/verification-cockpit" /path/to/my-app/.cursor/skills/Open that project in Cursor, then invoke the skill. It generates .agents/verification/* in that repo — tweak manifest and layout for that stack.
Prerequisite: av workflow with verify-launch.sh + verify-layout.sh on the host. See shell-av-workflow overlay in the skills library.
When to run
- Target project (or this shell repo as dogfood/stress test) needs a verification dashboard
AGENTS.md/ README lists verify commands not reflected in panes- Stack changed (new
pnpmscripts, Rust crate, CI jobs) - User asks for "verification cockpit", "av layout", or "mission control"
Workflow
Copy this checklist and track progress:
- [ ] 1. Discover verification commands
- [ ] 2. Classify launch tiers
- [ ] 3. Design pane layout
- [ ] 4. Write .agents/verification/*
- [ ] 5. Symlink .cursor/verify
- [ ] 6. Optional: AGENTS.md cockpit row1. Discover
Read in order (stop when you have enough signal):
| Source | Look for |
AGENTS.md | verify-before-done, post-change commands, stability hotspots |
docs/SETUP.md, README.md | build/test/lint commands |
package.json scripts | test, build, lint, dev |
Makefile, Cargo.toml, justfile | check/test targets |
.github/workflows/* | CI verify steps |
.agents/skills/*/SKILL.md | domain-specific verify |
2. Classify tiers
| Tier | Auto on av? | Examples |
monitor | yes | lazygit, empty console — omit yazi/btop unless they surface verify failures |
watch | yes | pnpm test --watch, cargo watch -x check, vitest --watch |
verify | confirm [y/N] | pnpm test, cargo test, pnpm build, tsc --noEmit |
mutate | blocked unless av --launch-mutate + type YES | pnpm install, migrations, deploy, format-all |
Rule: if it writes deps, data, or project structure → mutate. If it only reads/compiles/tests → verify or watch.
Set risk_profile in manifest: low | medium | high (from AGENTS.md stability contracts).
3. Design layout (two-pass, golden ratio)
Mandatory: every pane answers what failure does this surface? If it does not, omit it. Prefer four high-signal panes over six decorative ones.
Golden ratio: all splits use φ ≈ 1.618 → 62% major / 38% minor (bin/lib/verify-layout.sh). Nest splits so higher priority panes accumulate major shares.
Pass 1 — priority → area
Rank panes priority: 1 (highest) through N. Allocate area in golden proportions:
| Prio | Typical pane | Column / band |
| 1 | Primary watcher (test/lint/health watch) | Ops column — major height in right stack |
| 2 | CMD console | Ops column — minor height bottom-right (default focus) |
| 3 | GIT (lazygit) | Left column — major width (62%), full height |
| 4 | Verify-tier one-shot | Ops column — minor height top-right |
| 5+ | Second watcher / domain verify | Only if distinct failure signal; split ops stack again |
Use verify_layout_build_golden_grid from verify-layout.sh for the default 4-pane skeleton.
Pass 2 — context → arrangement
Adjust using space_profile per pane (manifest + reference.md):
| Profile | Output shape | Space rule |
scroll | streaming logs, test output | Largest vertical band in ops stack (right center) |
interactive | short commands, agent_scan | Compact bottom band (38% height), bottom-right |
tui-side | lazygit, tig | Major left column (62% width), full height |
confirm-burst | build/test on demand | Small bottom band; confirm before run |
omit | btop, yazi (default) | Do not include in verify window |
Default golden grid
+----------------------------+------------------+
| | VERIFY (minor) |
| GIT (tui-side, 62% w) |------------------|
| lazygit full height | WATCH (scroll) |
| |------------------|
| | CMD (interactive)|
+----------------------------+------------------+
git column 62% ops column 38%- CMD —
tier: monitor, no command —agent_scan,gdf,vf - WATCH — highest-priority watcher for this stack
- VERIFY — full-suite or build; confirm in pane
- Optional second window
verify-riskonly when many amber/red commands would crowd one window
Value audit (before shipping)
- [ ] Each pane has `value:` in manifest — one concrete failure mode
- [ ] No system monitors (btop) unless debugging perf during verify
- [ ] No file browser unless verify workflow inspects files
- [ ] No duplicate signals (two panes showing same test output)
- [ ] WATCH pane shows live output without manual refresh4. Write artifacts
Create under .agents/verification/:
| File | Purpose |
cockpit.yaml | Unified verify + test manifest (cockpits.verify, cockpits.test) |
manifest.yaml | Legacy verify pane map (still supported) |
tests.yaml | Legacy test runners (still supported) |
tmux-layout.sh | Executable layout (chmod +x) |
tmux-theme.conf | Optional project theme overrides |
README.md | Human pane legend |
Use templates from templates/ in this skill directory. Fill PROJECT, RISK, and PANES from discovery.
tmux-layout.sh contract:
- Args:
[directory](default.) - Must run inside tmux
- Sets
@workflow_dir,@workflow_mode verify - Idempotent:
verify_layout_ok— recreate when CMD missing or placeholder panes (FILES/SYS/INSIGHT/VERIFY) - Resolves project layout by walking up from cwd for
.agents/verification/tmux-layout.sh - Sources
verify-launch.sh+verify-layout.sh - Calls
verify_layout_build_golden_grid,verify_apply_theme,verify_launch_pane,verify_maybe_rescan - Ends with
tmux select-paneon console
5. Symlink
mkdir -p .cursor
ln -sfn ../.agents/verification .cursor/verify6. Optional AGENTS.md row
If no cockpit section exists, add under setup/verify:
| Verification cockpit | `.agents/verification/README.md` — run `av` in tmux after agent work |Test
In Ghostty + tmux (not Cursor integrated terminal):
t && z <project>
av # project layout; watchers auto-start
av --scan # + agent_scan in console
av --launch-mutate # allow mutate-tier confirms
av --generic # fallback to generic 4-pane cockpitReference
- Manifest schema: reference.md
- Starter templates: templates/ — copy into target project
.agents/verification/ - Runtime integration overlay: shell-av-workflow
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
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.

