5chan

Imported from tamarillo-ai/theta (test/cast-fixtures/codex-cli/regression/5chan/AGENTS.md). Install upstream with npx skills add tamarillo-ai/theta --skill 5chan. Copyright stays with the author.

AGENTS.md

Purpose

This file defines the always-on rules for AI agents working on 5chan. Use this as the default policy. Load linked playbooks only when their trigger condition applies.

Surprise Handling

The role of this file is to reduce recurring agent mistakes and confusion points in this repository. If you encounter something surprising or ambiguous while working, alert the developer immediately. After confirmation, add a concise entry to docs/agent-playbooks/known-surprises.md so future agents avoid the same issue. Only record items that are repo-specific, likely to recur, and have a concrete mitigation.

Project Overview

5chan is a serverless, adminless, decentralized imageboard built on the Bitsocial protocol.

Instruction Priority

• MUST rules are mandatory.
• SHOULD rules are strong defaults unless task context requires a different choice.
• If guidance conflicts, prefer: user request > MUST > SHOULD > playbooks.

Agent Operating Principles

• Before editing, state important assumptions when the task is ambiguous. Ask instead of silently choosing between materially different interpretations.
• Prefer the smallest implementation that solves the requested problem. Do not add speculative abstractions, configurability, or features.
• Keep diffs surgical. Do not refactor, reformat, rename, or "improve" adjacent code unless it is necessary for the task.
• Clean up only artifacts created by the current change, such as newly unused imports or dead helper code.
• For non-trivial work, define success criteria and verify them with the narrowest reliable checks before marking the task complete.

Task Router (Read First)

Stack

• React 19 + TypeScript
• Zustand for shared state
• React Router v6
• Vite
• bitsocial-react-hooks
• i18next
• yarn
• oxlint
• oxfmt
• tsgo

Project Structure

src/
├── components/   # Reusable UI components
├── views/        # Page-level route views
├── hooks/        # Custom hooks
├── stores/       # Zustand stores
├── lib/          # Utilities/helpers
└── data/         # Static data

Core MUST Rules

Package and Dependency Rules

• Use Corepack-managed Yarn 4, never npm. Run corepack enable once on a new machine before using yarn.
• Pin exact dependency versions (package@x.y.z), never ^ or ~.
• Keep lockfile synchronized when dependency manifests change.

React Architecture Rules

• Do not use useState for shared/global state. Use Zustand stores in src/stores/.
• Do not use useEffect for data fetching. Use bitsocial-react-hooks.
• Do not sync derived state with effects. Compute during render.
• Avoid copy-paste logic across components. Extract custom hooks in src/hooks/.
• Avoid boolean flag soup for complex flows; model state clearly in Zustand.
• Use React Router for navigation; no manual history manipulation.

Code Organization Rules

• Keep components focused; split large components.
• Follow DRY: shared UI in src/components/, shared logic in src/hooks/.
• Add comments for complex/non-obvious code; skip obvious comments.

Git Workflow Rules

• Keep master releasable. Do not treat master as a scratch branch.
• If the user asks for a reviewable feature/fix and the current branch is master, create a short-lived task branch before making code changes unless the user explicitly asks to work directly on master.
• Name short-lived AI task branches by intent under the Codex prefix: codex/feature/*, codex/fix/*, codex/docs/*, codex/chore/*.
• Open PRs from task branches into master so review bots can run against the actual change.
• Open PRs as ready for review, not draft. Draft PRs prevent CodeRabbit, Cursor Bugbot, and similar review bots from running.
• Prefer short-lived task branches over a long-lived develop branch unless the user explicitly asks for a staging branch workflow.
• Use worktrees only when parallel tasks need isolated checkouts. One active task branch per worktree.
• If a new task is unrelated to the currently checked out branch, do not stack it on that branch. Create a new worktree from master and create a separate short-lived task branch there.
• Always give a new worktree a descriptive name that reflects the task (e.g. fix-login-redirect, not wt1, tmp, feature, or a numbered slug), so it can be identified at a glance in a long list of worktrees. When using ./scripts/create-task-worktree.sh, the <slug> argument must be that descriptive name.
• Prefer ./scripts/create-task-worktree.sh <feature|fix|docs|chore> <slug> when you need a new task worktree and do not have a stronger repo-specific reason to create it manually.
• Treat branch and worktree as different things: the branch is the change set; the worktree is the checkout where that branch is worked on.
• For parallel unrelated tasks, give each task its own branch from master, its own worktree, and its own PR into master.
• After a reviewed branch is merged, prefer deleting it to keep branch drift and merge conflicts low.

Bug Investigation Rules

• For bug reports tied to a specific file/line, check relevant git history before any fix.
• Minimum sequence: git log --oneline or git blame first, then scoped git show for relevant commits.
• Full workflow: docs/agent-playbooks/bug-investigation.md.

Verification Rules

• Never mark work complete without verification.
• After code changes, run: yarn build, yarn lint, yarn type-check.
• After adding or changing tests, run yarn test.
• Do not commit or force-add local rebuild output. build/ is the main generated build output in this repo; remove or restore generated output directories after local verification before committing.
• After React UI logic changes, run: yarn doctor.
• Treat React Doctor output as actionable guidance; prioritize error then warning.
• For UI/visual changes, verify with playwright-cli across Chrome/Blink, Firefox/Gecko, and WebKit/Safari.
• Cover desktop and a mobile viewport flow in each browser engine when the change affects layout, touch behavior, or responsiveness.
• For browser automation and verification, default to a fresh isolated playwright-cli session for reproducibility.
• If the task depends on existing auth, cookies, extensions, open tabs, or another live browser state, explicitly confirm whether to use a fresh isolated session or the contributor's current browser session.
• Do not assume permission to drive the contributor's active personal browser session.
• The shared hook verification path is strict by default. Only set AGENT_VERIFY_MODE=advisory when you intentionally need signal from a broken tree without blocking the session.
• Use yarn test:coverage as an advisory check when expanding test coverage or auditing risky logic; do not invent a repo-wide coverage gate unless the user asks for one.
• If verification fails, fix and re-run until passing.

Tooling Constraints

• Use gh CLI for GitHub work (issues, PRs, actions, dependabot, projects, search).
• Do not use GitHub MCP.
• Do not use browser MCP servers (cursor-ide-browser, playwright-mcp, chrome MCP, etc.).
• Use playwright-cli for browser automation.
• If many MCP tools are present in context, warn user and suggest disabling unused MCPs.

AI Tooling Rules

• Treat .codex/, .cursor/, and .claude/ as repo-managed contributor tooling, not private scratch space.
• Keep equivalent workflow files aligned across all toolchains when their directories contain the same skill, hook, or agent.
• Keep shared behavior equivalent while preserving harness-specific models, config formats, hook entry points, and tool invocation syntax.
• Do not configure .claude agents to use composer-2; that model is Cursor-only in this repo. Keep .claude agent models on Claude-supported options.
• Do not configure .codex/agents/*.toml with gpt-5.3-codex or gpt-5.3-codex-spark; standardize Codex agents on gpt-5.4 unless the user explicitly requests a different model.
• When changing shared agent behavior, update the relevant files in .codex/skills/, .cursor/skills/, .claude/skills/, .codex/agents/, .cursor/agents/, .claude/agents/, .codex/hooks/, .cursor/hooks/, .claude/hooks/, and their hooks.json or config entry points as needed.
• If AGENTS.md references a skill, agent, or hook, prefer a tracked file under .codex/, .cursor/, or .claude/ rather than an untracked local-only instruction.
• Review .codex/config.toml, .cursor/hooks.json, and .claude/hooks.json before changing agent orchestration or hook behavior, because they are the entry points contributors will actually load.
• Before finishing any React UI logic change under src/components, src/views, src/hooks, or UI stores, review the changed diff with vercel-react-best-practices and, in Codex/Vercel-plugin sessions, vercel:react-best-practices. Fix valid findings before final verification; do not limit this review to diffs that add new hooks or memoization.
• When a diff adds new useEffect, useLayoutEffect, useInsertionEffect, useMemo, useCallback, or memo(...) usage under src/, treat the repo hook reminder as mandatory and also reconsider the change with you-might-not-need-an-effect before finishing.
• Directory-specific auto-loaded rules live under src/AGENTS.md and scripts/AGENTS.md; read them before editing files in those trees.
• For work expected to span multiple sessions, keep explicit task state in a feature-list.json plus progress.md pair using docs/agent-playbooks/long-running-agent-workflow.md.
• If more than one human or toolchain needs the same task state, keep it in a tracked location such as docs/agent-runs/<slug>/ instead of burying it in a tool-specific hidden directory.

Project Maintenance Rules

• If CHANGELOG.md or package version changes, run yarn blotter:check.
• If blotter check fails for missing release coverage, add a concise one-line release entry.
• Blotter entries are shown to 5chan end users, not developers. Every release message must be plain English a non-dev can understand: no internal library names (e.g. Pretext, Zustand), no dev shorthand (perf, deps, refactor, a11y, CI), no file paths, and no PR numbers. Prefer user-visible outcomes over implementation details (performant account history, not compact account history). Broad wording like security fix, stability fixes, or bug fixes is fine when the change is hard to summarize. Full rules and examples live in the release skill's Step 3.
• Ignore manual entries for release coverage logic.
• Use yarn blotter / yarn blotter:manual for manual dev messages.

Security and Boundaries

• Never commit secrets or API keys.
• Never push to a remote unless the user explicitly asks.
• Test responsive behavior on mobile viewport.

Core SHOULD Rules

• Keep context lean: delegate heavy/verbose tasks to subprocesses when available.
• For complex work, parallelize independent checks.
• Add or update tests for bug fixes and non-trivial logic changes when the code is reasonably testable.
• When touching already-covered code, prefer extending nearby tests so measured coverage does not regress without a clear reason.
• Use yarn knip when adding/removing dependencies or introducing new direct imports; treat findings as advisory, but resolve real issues before finishing.
• When proposing or implementing meaningful code changes, include both:
  • a Conventional Commit title suggestion
  • a short GitHub issue suggestion Use the format playbook: docs/agent-playbooks/commit-issue-format.md.
• When stuck on a bug, search the web for recent fixes/workarounds.
• After user corrections, identify root cause and apply the lesson in subsequent steps.

Local Development URLs

This project uses Portless for local dev. The canonical dev URL is https://5chan.localhost, and non-master branches can automatically fall back to a branch-scoped *.5chan.localhost route when needed so parallel worktrees do not collide. Other Bitsocial projects use the same proxy (seedit, mintpass, bitsocial at .localhost), so they can all run simultaneously without port conflicts. The launcher intentionally starts the Portless HTTPS proxy on port 443 before registering app routes; keep that step so legacy ~/.portless state on port 1355 is not reused.

To bypass Portless: PORTLESS=0 yarn start

Android phone over USB (default browser opens via adb; ANDROID_USB_OPEN_BROWSER=0 to skip): yarn start:android-usb

Common Commands

corepack yarn install
yarn start                # https://5chan.localhost
yarn start:android-usb    # Vite + adb reverse for USB Android (http://localhost:3000 on device)
yarn build
yarn test
yarn test:coverage
yarn knip
yarn knip:full
yarn prettier
yarn electron
yarn doctor
yarn doctor:score
yarn doctor:verbose
./scripts/create-task-worktree.sh chore ai-workflow-improvement
./scripts/agent-init.sh --smoke

Playbooks (Load On Demand)

Use these only when relevant to the active task:

• Hooks setup and scripts: docs/agent-playbooks/hooks-setup.md
• Long-running agent workflow: docs/agent-playbooks/long-running-agent-workflow.md
• Translations workflow: docs/agent-playbooks/translations.md
• Commit/issue output format: docs/agent-playbooks/commit-issue-format.md
• Skills/tools setup and MCP rationale: docs/agent-playbooks/skills-and-tools.md
• Bug investigation workflow: docs/agent-playbooks/bug-investigation.md
• Known surprises log: docs/agent-playbooks/known-surprises.md