FE-871: Brunch toolchain detection — detect + plan-emitter wiring by kostandinang · Pull Request #214 · hashintel/brunch

kostandinang · 2026-06-15T15:22:50Z

Stack Context

Stacks on FE-867 (#213). Second Arc-1 frontier -- closes the brunch detect gap FE-843 deferred. Slices 1 + 2 both land here.

What?

Resolve a real repo's toolchain to a registry ProfileId at plan time, and use it as the brownfield front of the FE-843 chain.

Slice 1 -- detector (detectProfile, project-detect.ts): pure, evidence-first -- the cheap "which lockfile/manifest is present" check, not a language-detection engine. Bun lockfile -> bun; deno config -> deno; package.json deps -> vitest/jest/none -> node-vitest/node-jest/node-test. One clear supported signal resolves; anything else fails loud ({detected:false, reason}), never a silent default; maps only to registry ids.
Slice 2 -- wire it in (resolveEmittedProfile, plan-emitter.ts): detected becomes the brownfield front of the chain --

--profile >> detected (brownfield) >> spec >> architect >> bun

repoDir threads CLI cwd -> runPlan -> emitPlanFromSnapshot; injectable detect seam keeps tests hermetic. A loud failure throws rather than silently falling to bun. Greenfield (no repoDir) keeps the unchanged FE-843 chain.

Why?

Per-slice verification runs one target at a time, and that command is runner-specific (bun test vs npx vitest run vs npx jest vs deno test). The toolchain must be known at plan time -- the deterministic runner reads the stamped plan.profile with no agent in the loop, so a wrong default runs the wrong command with no diagnostic. For an existing repo the manifest is the authoritative source; every other source is a guess or a manual step.

D160-K-clean: profile-id resolution is an input to authoring, not architect host-introspection -- no amendment needed.

Tests

detectProfile: bun lockfile -> bun, deno config -> deno, package.json deps -> vitest/jest/none -> the matching node profile; ambiguous evidence (vitest and jest) and non-JS/unknown repos fail loud with a reason -- never a silent default. resolveEmittedProfile: detected fronts the brownfield chain (over spec/architect/bun), --profile still overrides, a loud detection failure throws rather than falling to bun, and greenfield (no repoDir) keeps the unchanged FE-843 chain. Injectable detect seam keeps it hermetic.

Review fixes (this revision)

Bug -- ambiguous evidence was resolved by check-order. A package.json declaring both vitest and jest silently picked vitest, violating the fail-loud-not-silent-default invariant (acceptance Dolt #4). It now returns a loud {detected:false} reason and defers to --profile.
Overengineering trim. Dropped the pyproject.toml/setup.py/go.mod special-cases (+ the unsupported helper): they only renamed an error the catch-all already produces actionably -- the "language-detection framework" the design explicitly rejects. Non-JS repos still fail loudly via one actionable catch-all; only the per-stack naming is gone.

Co-authored-by: Amp amp@ampcode.com

kostandinang · 2026-06-15T15:23:10Z

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

This stack of pull requests is managed by Graphite. Learn more about stacking.

cursor · 2026-06-16T14:18:07Z

PR Summary

Medium Risk
Brownfield plans now depend on filesystem detection for stamped plan.profile and verification targets; mistakes could run the wrong runner, though ambiguous/unknown cases fail loudly and greenfield is explicitly no-op.

Overview
Brownfield brunch plan now infers the cook toolchain from the launch directory instead of defaulting to spec prose or bun. New detectProfile maps lockfiles/manifests to registry profiles (with monorepo workspace fallback when the root has no runner); ambiguous vitest+jest or unsupported stacks return {detected:false} rather than guessing.

The emitter inserts detection as --profile ≫ detected (brownfield only) ≫ spec ≫ architect ≫ bun via resolveEmittedProfile. Failed detection falls through to spec/architect when present, otherwise throws—no silent bun default. repoDir flows CLI cwd → runPlan → emitPlanFromSnapshot; injectable detect keeps tests hermetic.

Slice verification paths for brownfield co-locate with the host repo: detectTestDir samples existing *.test.* / *.spec.* layout (full dominant directory, including monorepo paths), and withTestDir rewrites profile default targets so narrowed vitest include globs still discover generated tests. Greenfield never runs detection or relocation.

Unit coverage spans project-detect, withTestDir, and emitter precedence / greenfield-invariant tests. memory/PLAN.md marks brunch-detect done.

^{Reviewed by Cursor Bugbot for commit a5eb05c. Bugbot is set up for automated code reviews on this repo. Configure here.}

cursor

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit f0c931a. Configure here.}

Add detectProfile(repoDir): a pure, evidence-first detector that maps a real repo's manifests/lockfiles to a registry ProfileId — the brownfield front of the FE-843 selection chain. Bun lockfile → bun; deno config/lock → deno; package.json vitest/jest/none → node-vitest/node-jest/node-test (lockfile evidence wins over package.json deps). Fails loudly instead of defaulting: known-but-unsupported stacks (Python via pyproject.toml/setup.py, Go via go.mod) and unrecognized dirs return {detected:false, reason} with an actionable message and the valid profile list — a wrong-but-silent toolchain produces unrunnable tests, the failure this closes. A malformed package.json is treated as a Node project (node-test). Pure detector only; wiring into the plan-emitter chain + the greenfield no-op and greenfield-protecting check are slice 2. Amp-Thread-ID: https://ampcode.com/threads/T-019ecb9a-9a08-733b-833d-76885fc8243a Co-authored-by: Amp <amp@ampcode.com>

Insert brunch-detect as the brownfield front of the FE-843 selection chain via resolveEmittedProfile: flag >> detected (brownfield) >> spec >> architect-classified >> bun. A loud detection failure throws rather than silently defaulting to bun (falling through to an explicit spec/architect choice first). Greenfield (or brownfield without a repoDir) keeps the unchanged chain. Thread repoDir from the CLI launch cwd through runPlan into emitPlanFromSnapshot; expose an injectable detect seam for hermetic emitter tests. Amp-Thread-ID: https://ampcode.com/threads/T-019ecb9a-9a08-733b-833d-76885fc8243a Co-authored-by: Amp <amp@ampcode.com>

… verification Fold the agent-native action vs harness-owned verification distinction into both frontier definitions: the value is the independent, deterministic, unshortcuttable check (the FE-800 orphan problem is unsolvable by agent self-report — same read-only discipline as evaluate-done at pi-actions.ts:70), not the boot/wiring action, which is agent-native via bash. Boot mechanics may lean on agent bash rather than a bespoke per-stack boot engine. Amp-Thread-ID: https://ampcode.com/threads/T-019ecb9a-9a08-733b-833d-76885fc8243a Co-authored-by: Amp <amp@ampcode.com>

…nches Review fix (oracle-confirmed): detectProfile silently resolved ambiguous evidence by check-order — a package.json declaring both vitest and jest picked vitest, violating the fail-loud-not-silent-default invariant (acceptance #4). It now returns a loud {detected:false} reason and defers to --profile. Also trims the pyproject.toml/setup.py/go.mod special-casing (+ the 'unsupported' helper): those only renamed an error the catch-all already produces actionably, which is the 'language-detection framework' the design explicitly rejects. Any repo with no JS/TS evidence now falls to one actionable catch-all. Behavior for non-JS repos is unchanged (still a loud, actionable failure); only the per-stack naming is gone. PLAN.md brunch-detect definition updated to match. Amp-Thread-ID: https://ampcode.com/threads/T-019ecb9a-9a08-733b-833d-76885fc8243a Co-authored-by: Amp <amp@ampcode.com>

Amp-Thread-ID: https://ampcode.com/threads/T-019ecb9a-9a08-733b-833d-76885fc8243a Co-authored-by: Amp <amp@ampcode.com>

Co-authored-by: Cursor <cursoragent@cursor.com>

This was referenced Jun 15, 2026

FE-843: Expand cook toolchain profiles, selected at plan time #198

Open

FE-864: Orchestrator improvements umbrella — brownfield feature delivery from spec #212

Open

kostandinang mentioned this pull request Jun 15, 2026

FE-867: Agent extension host — dual-mode pi-harness contract #213

Open

kostandinang changed the title ~~FE-871: brunch toolchain detection — detectProfile (slice 1)~~ FE-871: brunch toolchain detection — detect + plan-emitter wiring Jun 15, 2026

This was referenced Jun 15, 2026

FE-872: Install-failure classification — infra-vs-test split #218

Open

FE-875: App runtime probe — boot, probe, classify #219

Open

FE-876: Integration oracle — reachability gate + grounding seam #220

Open

kostandinang changed the title ~~FE-871: brunch toolchain detection — detect + plan-emitter wiring~~ FE-871: Brunch toolchain detection — detect + plan-emitter wiring Jun 16, 2026

This was referenced Jun 16, 2026

FE-877: Brownfield promotion — commit the cook result onto cook/<runId> #221

Open

FE-878: Brunch serve — one-shot plan-then-cook capstone #222

Open

kostandinang force-pushed the ka/fe-867-agent-extension-host branch from 0a176f5 to bcccd55 Compare June 16, 2026 12:53

kostandinang force-pushed the ka/fe-871-brunch-detect branch from c949e53 to a1afc98 Compare June 16, 2026 12:53

kostandinang mentioned this pull request Jun 16, 2026

FE-879: Lazy per-slice cook worktrees and shared node_modules for brownfield #223

Open

kostandinang marked this pull request as ready for review June 16, 2026 14:18

kostandinang mentioned this pull request Jun 16, 2026

FE-883: Orchestrator operational improvements (serve reliability + Opus 4.8) #224

Open

kostandinang force-pushed the ka/fe-867-agent-extension-host branch from bcccd55 to 86df4fc Compare June 16, 2026 23:45

kostandinang force-pushed the ka/fe-871-brunch-detect branch 2 times, most recently from 625b1b5 to f0c931a Compare June 16, 2026 23:55

kostandinang force-pushed the ka/fe-867-agent-extension-host branch from 86df4fc to 4cac54a Compare June 16, 2026 23:55

cursor Bot reviewed Jun 17, 2026

View reviewed changes

Comment thread src/orchestrator/src/project-detect.ts

kostandinang and others added 7 commits June 17, 2026 09:36

FE-871: co-locate generated tests in the repo's own test dir (slice 3)

6f38119

Amp-Thread-ID: https://ampcode.com/threads/T-019ecb9a-9a08-733b-833d-76885fc8243a Co-authored-by: Amp <amp@ampcode.com>

FE-871: monorepo-robust test-dir + workspace runner detection (slice 4)

bd84d85

Amp-Thread-ID: https://ampcode.com/threads/T-019ecb9a-9a08-733b-833d-76885fc8243a Co-authored-by: Amp <amp@ampcode.com>

FE-871: detect root-level test layouts

a5eb05c

Co-authored-by: Cursor <cursoragent@cursor.com>

kostandinang force-pushed the ka/fe-871-brunch-detect branch from f0c931a to a5eb05c Compare June 17, 2026 08:51

kostandinang force-pushed the ka/fe-867-agent-extension-host branch from 4cac54a to 7ae1e4f Compare June 17, 2026 08:51

kostandinang mentioned this pull request Jun 17, 2026

FE-881: Cook agent loads the target repo's sandbox-scoped skills #227

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

FE-871: Brunch toolchain detection — detect + plan-emitter wiring#214

FE-871: Brunch toolchain detection — detect + plan-emitter wiring#214
kostandinang wants to merge 7 commits into
ka/fe-867-agent-extension-hostfrom
ka/fe-871-brunch-detect

kostandinang commented Jun 15, 2026 •

edited

Loading

Uh oh!

kostandinang commented Jun 15, 2026 •

edited

Loading

Uh oh!

cursor Bot commented Jun 16, 2026 •

edited

Loading

Uh oh!

cursor Bot left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

kostandinang commented Jun 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Stack Context

What?

Why?

Tests

Review fixes (this revision)

Uh oh!

kostandinang commented Jun 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cursor Bot commented Jun 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR Summary

Uh oh!

cursor Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

kostandinang commented Jun 15, 2026 •

edited

Loading

kostandinang commented Jun 15, 2026 •

edited

Loading

cursor Bot commented Jun 16, 2026 •

edited

Loading