Help us improve
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
By krzemienski
Deepest-mode planning — consensus + gates + phase hierarchy + multi-plan tournament synthesis. Self-contained. Works with Claude Code and OpenCode.
npx claudepluginhub krzemienski/deepest-plan-plugin --plugin deepest-planOne-time deepest-plan validation init for this project (creates .planning/ + e2e-evidence/ + .gitignore entries).
Read-only validation audit — capture evidence and classify findings without changing code.
Non-interactive validation for CI/CD — no approval gates, exit codes, no fix loop.
Auto-fix validation failures and re-validate (3-strike protocol).
Generate e2e-evidence/validation-plan.md with PASS criteria for every journey (no execution).
WCAG 2.1 AA audit via axe-core, keyboard-navigation probe, and screen-reader label verification. Required for any UI plan that ships or modifies a user-visible surface. Optional skill listed in deepest-plan tool_enrichment when in scope.
API platform validation via curl / httpie against real running servers. Captures full response bodies, headers, status codes. Tests CRUD, auth, error paths, pagination, rate limiting. Cited as platform_skill on API gates in deepest-plan output.
Capture "before" state BEFORE making changes. Validation later compares against this baseline to prove improvement without regression. Invoked when deepest-plan modifies existing behavior (not greenfield).
CLI platform validation via direct binary execution. Captures stdout, stderr, exit codes, file output, pipe behavior. Cited as platform_skill on CLI gates in deepest-plan output.
Smart waiting patterns for async operations. Wait for observable conditions (DOM state, HTTP status, file presence, log line) with timeout and failure path — never arbitrary sleeps. Used inside deepest-plan preflight and validation_gate execute blocks.
Share bugs, ideas, or general feedback.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Multi-model consensus engine integrating OpenAI Codex CLI, Gemini CLI, and Claude CLI for collaborative code review and problem-solving.
Comprehensive UI/UX design plugin for mobile (iOS, Android, React Native) and web applications with design systems, accessibility, and modern patterns
Standalone image generation plugin using Nano Banana MCP server. Generates and edits images, icons, diagrams, patterns, and visual assets via Gemini image models. No Gemini CLI dependency required.
Write feature specs, plan roadmaps, and synthesize user research faster. Keep stakeholders updated and stay ahead of the competitive landscape.
Shannon Framework v6 — single-plugin consolidation replacing OMC + VF + Crucible + 13 others. 26 commands, 22 skills, 10 agents, 14 hooks across 7 domain modules + 4 enforcement layers.
Sharp-eyed visual-audit suite for Claude Code. Two coupled skills catch real UI defects — contrast failures, false affordances, modal opacity, contract mismatches — via real-system probes, zero mocks, evidence-cited verdicts.
Fixed-Point Deepen architecture of the Anneal plugin family. One plan, heated and cooled repeatedly — inline red team at every depth, Momus 0-100 scoring, convergence by variance/delta/cap.
3-agent unanimous consensus validation with hard gates for Claude Code. Maps Lead/Alpha/Bravo roles to CC subagents, enforces unanimity at phase transitions, persists evidence per phase+role.
Evidence-gated task planning, execution, and validation for Claude Code. Refuses completion without quorum-approved proof. No mocks. No stubs. No silent retries past the gate.
A self-contained Claude Code + OpenCode plugin for producing bulletproof implementation plans.
Generates consensus-validated, research-grounded, gate-hardened, phase-structured execution plans. Optional multi-plan tournament synthesis selects the best of N competing plans. Every plan ships with blocking validation gates that enforce real-system proof — no mocks, no stubs, no test files.
/deepest <task> — runs the full 6-phase pipeline/deepest:setup, /deepest:validate-plan, /deepest:validate, /deepest:validate-fix, /deepest:validate-audit, /deepest:validate-ci.planning/ tree OR single monolith XML OR bothDifferent runtime, different install path.
/plugin marketplace)The simplest path — install directly from GitHub:
/plugin marketplace add krzemienski/deepest-plan-plugin
/plugin install deepest-plan@deepest-plan-dev
Or clone locally first (useful for development / offline use):
git clone https://github.com/krzemienski/deepest-plan-plugin.git ~/Desktop/deepest-plan
cd ~/Desktop/deepest-plan
./scripts/install.sh claude # prints the two /plugin commands to run
Then inside a Claude Code session:
/plugin marketplace add ~/Desktop/deepest-plan
/plugin install deepest-plan@deepest-plan-dev
The repo doubles as its own marketplace — .claude-plugin/marketplace.json lists the plugin under the deepest-plan-dev marketplace name. Restart Claude Code to pick up commands and skills.
OpenCode plugins are code modules (JS/TS), not declarative manifests. But OpenCode ALSO autoloads skills and slash commands from ~/.config/opencode/{skills,commands}/. The installer symlinks the plugin's skills and commands into those dirs so every /deepest:* command and bundled skill is available in OpenCode sessions.
./scripts/install.sh opencode
Override target via OPENCODE_CONFIG_DIR env var.
./scripts/install.sh both
python3 scripts/validate-plugin.py
Expected: VALIDATION PASSED.
Inside a fresh session:
/help should list the /deepest commands/deepest:setup should autocomplete./scripts/uninstall.sh both
For Claude Code you still need /plugin uninstall deepest-plan@deepest-plan-dev inside a session; the script prints the command.
# In any project
/deepest Add a /account/preferences page where users toggle email preferences
The plugin will:
Output: .planning/ tree with BRIEF.md, ROADMAP.md, and per-phase PLAN.md files, plus e2e-evidence/ scaffolded for execution.
Then to execute:
/deepest:setup # one-time init
/deepest:validate-plan # generate validation-plan.md
/deepest:validate # run the full pipeline
More: docs/QUICKSTART.md.
| Mode | When to use | Flag |
|---|---|---|
tree (default) | Multi-phase projects, team review, PR-oriented | --tree |
monolith | Single XML mega-prompt for autonomous one-shot execution | --monolith |
both | Humans read the tree, agents consume the monolith | --both |
Competitive plan generation. Launches N parallel plan generators, a tailored meta-judge eval spec, M parallel judges, then routes adaptively:
Invoke:
/deepest --synthesis 5 "Refactor auth" # tournament of 5
/deepest --synthesis 5 --debate "Migrate payments" # debate judges
Detail: skills/deepest-plan/references/synthesis-workflow.md and synthesis-config.md.
Every skill this plugin references is bundled inside this plugin. Nothing external required.