Skip to content

chore: align CI to the hardened canon, add AGENTS.md#11

Merged
Unisay merged 1 commit into
masterfrom
chore/agents-md-and-canon-ci
Jun 14, 2026
Merged

chore: align CI to the hardened canon, add AGENTS.md#11
Unisay merged 1 commit into
masterfrom
chore/agents-md-and-canon-ci

Conversation

@Unisay

@Unisay Unisay commented Jun 14, 2026

Copy link
Copy Markdown
Collaborator

Brings this fork in line with the shared tooling canon and gives the AI coding agents a single source of truth.

CI (.github/workflows/ci.yml):

  • Drop accept-flake-config = true. It let a PR's own flake nixConfig inject extra substituters and keys, which is a supply-chain risk; the caches are already pinned in extra_nix_config, so nothing is lost.
  • luacheck now runs --std lua51 --no-unused-args instead of --std min. lua51 matches the actual target (it knows math.pow/math.atan2, which min flags), and --no-unused-args tolerates the curried fallback args the native FFI stubs ignore.
  • Run the test step as bash ./scripts/test so it no longer depends on the file's execute bit.

AGENTS.md + CLAUDE.md:

  • AGENTS.md is the one instruction file the coding agents read natively (Codex, Cursor, Copilot, Gemini CLI, Aider): the commands, the Lua 5.1 constraints, the FFI parenthesisation rule, the toolchain pins, and a pointer to the package-set ADRs.
  • CLAUDE.md is a one-line @AGENTS.md import so Claude Code reads the same file instead of a second copy that drifts out of sync.

It also drops the dead .eslintrc.json, trims the stale JavaScript entries from .gitignore, and bumps the pslua pin to current main (the purescript#56/purescript#49/purescript#61 fixes).

Tooling and docs only, no src/ change, so there is no tag or package-set bump.

Verification

nix develop -c ./scripts/build compiles the fork on the overlay shell and luacheck with --std lua51 --no-unused-args src/ is clean locally (17 files).

CI: drop accept-flake-config, luacheck --std lua51 --no-unused-args, test via
bash. Add AGENTS.md (single agent instruction file) + CLAUDE.md @-import.
Clean dead JS-era files and .gitignore entries. Bump pslua pin to current
main. Tooling and docs only, no src/ change.

Copilot AI left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

Aligns this fork’s CI and agent-facing documentation with the shared “hardened canon”, reducing Nix supply-chain risk and centralizing AI agent instructions.

Changes:

  • Harden Nix CI by removing accept-flake-config = true and make the test step independent of the executable bit.
  • Update luacheck invocation to match the Lua 5.1 target and tolerate intentionally-unused stub arguments.
  • Add AGENTS.md as the single source of truth for coding-agent guidance, with CLAUDE.md importing it; remove dead ESLint config and trim stale JS ignores.

Reviewed changes

Copilot reviewed 4 out of 6 changed files in this pull request and generated no comments.

Show a summary per file
File Description
flake.lock Bumps the pslua (Unisay/purescript-lua) pin to a newer revision.
CLAUDE.md Imports AGENTS.md so Claude Code reads the same unified agent instructions.
AGENTS.md Adds consolidated dev commands + Lua 5.1/FFI/toolchain/release guidance for coding agents.
.gitignore Removes now-stale JS-related ignore entries and the unignore for the deleted .eslintrc.json.
.github/workflows/ci.yml Hardens Nix config handling and adjusts test + luacheck commands to match documented workflow.
.eslintrc.json Removes an unused ESLint configuration file.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

@Unisay Unisay merged commit 78185b3 into master Jun 14, 2026
2 checks passed
@Unisay Unisay deleted the chore/agents-md-and-canon-ci branch June 14, 2026 17:43
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants