Skip to main content
rule CX Codex

parallax — Claude Code rule: workflow and installation

The parallax Claude Code rule structures development around discussion, a reviewed plan, and independent checks. Work then moves to a clean session for implementation and reconciliation.

MIT ★ 136 updated 2026-11-16

Install into GitHub Copilot · Manually: put the files in place All methods and platforms ↓
.github/instructions/parallax.instructions.md
Open on GitHub

What parallax does

The rule sets out a development flow from initial discussion to a final review. qstart aligns the discussion; after qgo, the agent prepares a plan for the independent bcheck review. A score of at least 4 is required to pass.

Once approved, gclear hands the task to a clean session. That session implements and verifies the work, then rcheck compares the result with the plan; it also needs a score of at least 4. The sequence ends with a scoped local commit and xwrap.

A subagent in an independent context handles each review, so reviewing the work in the same session is forbidden. Discussion notes, plans, reports, and hand-off material go into the relevant docs/dev/ folders. Automatic hand-off and reviews rely on session and delegation tools. Manual permission works too, but the first call to each tool needs approval.

What it helps with

  • align on a development direction before drafting a plan
  • get an independent review of the plan before hand-off
  • move the task and its context into a clean session
  • compare implementation with the plan before a local commit

Example prompts for the agent

  • Start a development discussion for this task with qstart.
  • The plan is ready; pass the work to a new session with qgo.
  • Compare the implementation with the plan, then finish with rcheck and xwrap.

How to install parallax

Manually: put the files in place

.github/instructions/parallax.instructions.md

The applyTo field sets which files it applies to

Step by step, with commands for each platform: AGENTS.md vs CLAUDE.md: what they are and which one to use

Questions about parallax

How do I install parallax in Claude Code?

Run these commands in order: /plugin marketplace add AI4XE/Parallax, then /plugin install parallax@parallax. The rule file is at .claude/rules/parallax.md.

What does the parallax rule do?

It organizes development into discussion, an independently reviewed plan, a hand-off to a clean session, implementation checks, and wrap-up.

How does parallax review work?

The bcheck subagent reviews the plan, and rcheck reconciles the implementation. Both reviews use an independent context and require a score of at least 4 to pass.

Source text

parallax.instructions.md · MIT
---
description: Parallax development flow (suite instructions)
applyTo: "**"
---
# Parallax development flow (suite global instructions)

This VS Code installation has the Parallax lightweight development flow suite available (bundled with the Parallax extension, solo direct-execution tier): structured discussion → reviewed plan → independent review gate → automatic hand-off to a clean session → reconciliation wrap-up. The seven skills (including wish) are provided by this extension and can be invoked explicitly (`/skill-name`) or triggered automatically from natural language.

## The flow in one sentence

qstart (discussion alignment) → the user says qgo → plan + independent bcheck review (PASS≥4) → gclear automatic hand-off to a clean session → hand-off session executes + real verification → rcheck reconciliation (PASS≥4) → scoped local commit → xwrap wrap-up.

## Skill trigger words

| Skill | Trigger phrases (examples) |
|---|---|
| qstart | "qstart", "Q development discussion", "start a discussion" (aliases ver/version→version task, nover→non-version task) |
| qgo | "qgo", "Q hand-off development", "the direction is fine, hand it off" |
| bcheck | "bcheck", "plan review", "analysis review" |
| gclear | "gclear", "clear and hand off", "hand off to a new session", "cross-session context hand-off" |
| rcheck | "rcheck", "reconciliation review", "wrap-up review" |
| xwrap | "xwrap", "wrap up", "wrap up now" |
| wish | "wish", "许愿创造", "许愿编程", "Parallax" |

## GATE marker semantics

- `GATE:bcheck:PASS` / `GATE:bcheck:FAIL:{score}`: plan review gate; only a score ≥4 passes.
- `GATE:rcheck:PASS` / `GATE:rcheck:FAIL:{score}`: implementation reconciliation gate; same as above.
- `HANDOFF-ACK:<token>`: gclear hand-off receipt — the new session's first reply after receiving the hand-off prompt must carry this marker (the token matches the hand-off material).
- Gate reviews are always performed by an independent-context subagent (bcheck/rcheck delegate via task); same-session self-review is forbidden; `write_agent` is not used to create agents at runtime.

## Artifact locations

Discussion → `docs/dev/infra/` or `docs/dev/hotfix/`; analysis/plan → `docs/dev/analysis/`; review reports → `docs/dev/review/` or `docs/dev/analysis/`; hand-off material → `docs/dev/handoff/`. Follow the project's existing document conventions if any.

## Recommended setting (one-time, optional)

The skill chain's automatic hand-off (gclear) and automatic review depend on the session tools and delegation tool. It is recommended to set the chat permission tier to **zero-confirmation fully automatic** (setting key `chat.defaultConfiguration`, choose Bypass Approvals or Autopilot) — cross-session messages and delegation then have no pop-ups and the whole chain needs no manual steps. The default Manual tier also works: each tool's first call is approved once and same-kind calls in the same session are then auto-exempt, with just a few more clicks.

© AI4XE · MIT · AI4XE/Parallax

Similar IDE rules

IDE rules catalog →
moral-parallax jwynia moral-parallax generates speculative fiction narratives about systemic exploitation by closing the moral distance between actions and harms. Best for stories where innocence is impossible. CX Codex +1 428installs scrollytelling-and-parallax-data-visualization official OpenAI Guides the design and implementation of scrollytelling and parallax data visualizations. Use it when scroll progression explains data through staged scene reveals. CX Codex GP ChatGPT +1 13installs parallax-landing-page hoodini parallax-landing-page — a Claude Code skill that turns a short video into a cinematic landing page: frames scrub with scroll while the document stays fixed. Text overlays shift as scroll progress advances. CX Codex +1 ★ 301repo stars context-mode Mert Koseoğlu context-mode routes heavy tool outputs and web pages into an MCP sandbox so raw data does not clutter the agent context window. CX Codex +1 ★ 34,370repo stars context-mode Mert Koseoğlu This rule routes data work through context-mode MCP tools and keeps raw files out of the conversation. Use it for analysis, search, and transformations. CX Codex +1 ★ 34,370repo stars json-structured-prompting calesthio json-structured-prompting defines a JSON architecture for complex image prompts with multiple subjects, explicit positioning, and variable templates. CX Codex +1 ★ 78,316repo stars i2i-prompting calesthio i2i-prompting — a rule for writing image-to-image editing prompts with FLUX.2. Use it for targeted changes, style transfers, text edits, and staged transformations. CX Codex +1 ★ 78,316repo stars flux2-models calesthio Rule with prompting patterns for FLUX.2 models: klein, max, pro, flex, and dev. Shows the agent how to write generation and editing prompts for each variant. CX Codex +1 ★ 78,316repo stars

From the same set19

The whole set →
parallax-reviewer The parallax-reviewer Claude Code subagent independently checks plans and implementation claims. Its report ends with a GATE:PASS or GATE:FAIL marker and a score. CX Codex +1 ★ 136repo stars qgo This Claude Code skill reviews a task plan, hands it to a fresh session, and guides execution through verification. It activates on qgo or an explicit hand-off in a Q discussion. CX Codex +1 ★ 136repo stars qstart The qstart Claude Code skill takes a task from requirement intake through research and decision alignment. It waits for an explicit qgo before handing the task off for execution. CX Codex +1 ★ 136repo stars gclear The gclear Claude Code skill carries task context into a fresh session and sends it instructions to continue. It starts after planning or discussion when work needs to move to a new session. CX Codex +1 ★ 136repo stars gclear The gclear Claude Code skill carries working context into a fresh session after clearing the current one. It runs when a task needs to continue in a clean conversation. CX Codex +1 ★ 136repo stars qstart qstart — a Claude Code skill for gathering requirements and working through a development task. It waits for an explicit qgo before handing the agreed plan to execution. CX Codex +1 ★ 136repo stars gclear gclear — a skill for Claude Code hands work into a refreshed model context in the same conversation. It activates when a long discussion needs a compact state and continued execution. CX Codex +1 ★ 136repo stars rcheck rcheck — a skill for Claude Code compares finished work with its plan and implementation summary. It runs after a complex task to check completion and supporting evidence. CX Codex +1 ★ 136repo stars

Project sponsors

Companies that keep this analytics open