Skip to content

Set up Cursor Cloud development environment - #6

Draft
carenthomas wants to merge 2 commits into
mainfrom
caren/dev-environment-setup-6203
Draft

Set up Cursor Cloud development environment#6
carenthomas wants to merge 2 commits into
mainfrom
caren/dev-environment-setup-6203

Conversation

@carenthomas

@carenthomas carenthomas commented Jul 20, 2026

Copy link
Copy Markdown
Contributor

Summary

Sets up the development environment for the trajectory library (TypeScript core + Python wrapper) in Cursor Cloud and documents durable operating guidance for future agents.

Changes

  • Add AGENTS.md with a ## Cursor Cloud specific instructions section covering the toolchain (Bun 1.3.0, Node 20+, Python 3.10+), the main commands, and non-obvious gotchas (regenerating the vendored Python CLI bundle after TS edits; Deep Agents tests skipping without optional extras).
  • Add a ## Canonical-contract work: invariants and review learnings section capturing the chunk-stable identity / per-adapter anchor / partial-continuation invariants for normalizeToCanonical, plus a process note to red-team invariants with the design authority and self-audit per-adapter before requesting review.

Environment setup

The startup update script installs Bun 1.3.0 (if missing) and runs bun install --frozen-lockfile. No code changes were required — the repo was already working.

Verification

  • bun run check — typecheck + tests + build.
  • PYTHONPATH=python/src python3 -m unittest discover -s python/tests -v — Python parity.
Open in Web Open in Cursor 

cursoragent and others added 2 commits July 20, 2026 17:11
Co-authored-by: cthomas <carenthomas@users.noreply.github.com>
Co-authored-by: cthomas <carenthomas@users.noreply.github.com>

@cpacker cpacker left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

The canonical-contract guidance currently contradicts the implementation and should be corrected before this becomes repository-wide agent guidance:

  • Id-less Letta Code rows use ordinal line anchors, not byte anchors.
  • normalizeTranscript() does enable partial semantics when baseByteOffset is nonzero; it does not simply “stay strict.”
  • Meta omission for continuation chunks is a canonical-projection rule and should not be conflated with general transcript normalization.

Because future coding agents will treat AGENTS.md as authoritative, these inaccuracies are more harmful than ordinary documentation drift. Please derive this section directly from CANONICAL.md and the current adapter code. Also replace the internal downstream repository name with generic public-safe wording.

👾 Generated with Letta Code

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.

3 participants