| name | ultrathink-deep-dive-loop |
| description | Full deep-dive audit loop — bugs, edge cases, backlog, current competitive web research, TDD fix waves, council-reviewed plans, re-audit until clean. Use for ultrathink mandates, /deep-dive, security audits, complete backlog, competitive analysis. Routes orchestrator by provider (Claude Opus vs Fable, Codex gpt-5.5, Cursor Auto vs Opus, agy Gemini 3.5/3.1). Spawns workflows + subagents; a peer-review council audits plans. |
ultrathink-deep-dive-loop
Provider-agnostic procedure. Read the appendix for your active CLI before Phase 0.
Trigger phrases: ultrathink, deep dive, complete backlog, security audit,
competitive analysis, /deep-dive, /security-audit, /audit-loop.
Phase 0 — Route orchestrator
Classify the mandate:
- Security / audit / vuln / money / auth → top-tier orchestrator, security council (no Fable on Claude).
- Feature / design / writing → Fable OK on Claude; Codex gpt-5.5; Cursor Auto OK for mechanical.
- Mechanical fix-loop → mid-tier orchestrator + cheap scan seats.
Read core/prompts/principles/09-orchestrator-routing.md.
Phase 1 — Scan (cheap seats)
- Local inventory —
tg doctor, tg map or tg orient, scoped tg search.
- Backlog — open issues, TODO/FIXME, failing CI,
docs/ audit queues.
- External truth — available web-research tools and authoritative current sources for:
- competitor / prior-art (T-72h freshness)
- library behavior you would otherwise trust from training data
- novel edge-case ideas for this product surface
- Competitive pass — document 3+ alternatives + differentiation + untested edge cases.
Cap fan-out. Log dropped agents (no silent caps).
Phase 2 — Plan (top-tier orchestrator)
- Load superpowers:writing-plans (install superpowers plugin if absent).
- TDD-shaped plan: failing test → fix → green per item.
- Decide workflow vs inline subagents:
- >3 independent files OR cross-verify →
Workflow tool
- Template:
core/workflows/deep-dive-bughunt.js (pass repo, root, lenses)
- Council audit the plan (before any code):
- Security: a peer-review council + codex (+ sonnet-5 seat; no Fable)
- Else: Fable + codex + council
- Background council →
CronCreate follow-up per the council mandate
- Risky fixes → a fix-approach review pass on proposed fixes (not findings).
Phase 3 — Build (mid-tier, one-writer-per-repo)
- sonnet / gpt-5.5-medium / Cursor Auto+inherit for implementation subagents.
- Verify subagent output (receipts, commit, tests) on every multi-phase dispatch.
- Large reads → gotcontext MCP; local symbols →
tg callers / tg blast-radius.
- agy seat for parallel read-only audit lenses (bulk: 3.5 Flash High).
Phase 4 — Review diff
- Security: codex + thinktank + sonnet-5 (no Fable).
- Else: Fable + codex (+ thinktank on load-bearing).
- Merge gate (Opus / gpt-5.5-high): verify EACH finding at
file:line yourself.
- Full suite green + lint + format.
- Engine/API/MCP change → dogfood live gotcontext MCP.
Phase 5 — Re-audit loop
Until clean:
/audit-loop or re-run deep-dive-bughunt on changed paths.- Competitive edge-case retest (current authoritative web research if external).
- Stop only with evidence — not self-report.
Workflow: deep-dive-bughunt
Workflow({
name: 'deep-dive-bughunt',
args: {
repo: 'my-app',
root: '/path/to/repo',
lenses: [
{ key: 'api', focus: 'api/ handlers auth billing', maxFiles: 12 },
{ key: 'web', focus: 'frontend routes + hooks', maxFiles: 12 },
],
},
})
Output: ranked fix queue → fix-approach review → TDD waves.
Appendix: Claude
| Phase | Model | Notes |
|---|
| Orchestrator security | /model opus persist | NEVER Fable |
| Orchestrator feature | Fable OK | |
| Scan | haiku Agent | |
| Build | sonnet Agent worktree isolation | |
| Council | thinktank + codex | Security: drop Fable |
| Merge | Opus | Always |
Commands: /deep-dive, /security-audit, /audit-loop, /competitive-exa
Hooks: ultrathink_orchestrator_gate injects routing on trigger phrases.
Appendix: Codex
| Phase | Model | Command |
|---|
| Orchestrator | gpt-5.5 | model = "gpt-5.5" in config.toml |
| Plan effort | high | model_reasoning_effort = "high" |
| Scan subagent | gpt-5.4-mini | codex exec -m gpt-5.4-mini -p "..." |
| Build | gpt-5.5 medium | model_reasoning_effort = "medium" |
| Cross-review | Claude Opus CLI | separate CLI session / claude -p |
| Gemini seat | agy | conpty-run.exe + use-gemini |
Profiles: copy providers/codex/profiles/security.toml vs feature.toml.
No hooks — you are the enforcement. MCP via
npx --yes gotcontext@1.0.1 wrap codex.
Prompt macro: @deep-dive from ~/.codex/prompts/deep-dive.md (after install).
Appendix: Cursor
| Phase | Model | Setting |
|---|
| Default mechanical | Auto | Composer subagents — cheap |
| Security orchestrator | Claude Opus 4.6 explicit | Enable Max Mode |
| Subagents | model: inherit | .cursor/agents/*.md |
| External seats | codex exec + agy | Parallel verification |
| Merge | Opus or pinned gpt-5.5 | Not Auto |
Auto trap: Auto picks Composer 2.x; fine for explore/bash, wrong for security synthesis.
Pin Opus for ultrathink security; Max Mode so subagents don't silently downgrade.
Commands: .cursor/commands/deep-dive.md (project) or ~/.cursor/commands/.
Hook: beforeSubmitPrompt → ultrathink_prompt.sh (after cursor install).
Tri-stack pattern: Cursor builds → codex reviews diff → agy third opinion on claims.
Appendix: agy (Antigravity CLI)
When agy is the active orchestrator (not just a third seat):
| Phase | Model | Command / path |
|---|
| Orchestrator (bulk) | Gemini 3.5 Flash (High) | agy --model "Gemini 3.5 Flash (High)" |
| Orchestrator (security) | Gemini 3.1 Pro (High) | agy --model "Gemini 3.1 Pro (High)" |
| Parallel scan lenses | 3.5 Flash (High) | saddle-agy-headless.sh (ConPTY) |
| Cross-review | codex gpt-5.5 + Claude Opus | codex CLI / a separate CLI session |
| Merge / synthesis | 3.1 Pro or defer to Opus | agy alone does not auto-merge security |
Profiles: providers/agy/profiles/security.json vs feature.json →
~/.gemini/profiles/ (copy model into settings.json).
No native hooks — GEMINI.md + ~/.gemini/prompts/ are binding law.
MCP: ~/.gemini/config/mcp_config.json (remote uses serverUrl). Verify: /mcp.
Prompt macros (after ./install.sh agy):
~/.gemini/prompts/deep-dive.md~/.gemini/prompts/security-audit.md~/.gemini/prompts/audit-loop.md
Read-only audits: agy --sandbox -p "..." + spec under .agents/tasks/.
When agy is third seat in Claude/Codex/Cursor sessions, keep the primary
orchestrator on Opus/gpt-5.5; agy runs parallel read lenses only.
Cross-provider agy dispatch
All providers share the same headless path:
export SADDLE_HOME=/path/to/gotcontext-saddle
"$SADDLE_HOME/providers/agy/bin/saddle-agy-headless.sh" \
--model "Gemini 3.5 Flash (High)" \
--add-dir /path/to/repo \
"Your audit prompt"
Windows direct (if conpty-run.exe on PATH):
conpty-run.exe "$HOME/AppData/Local/agy/bin/agy.exe" -p "PROMPT" \
--model "Gemini 3.5 Flash (High)" --print-timeout 120s
See use-gemini for WSL tmux fallback and council seat wiring.
Skills index (load on demand)
| Skill | Phase |
|---|
| Available web-research tools | 1, 5 competitive |
tensor-grep-code-audit | 1, 3 local code |
use-gotcontext | 3 compression |
use-gemini | 3–4 agy seats |
Superpowers plugin: writing-plans, test-driven-development, verification-before-completion.
