Skip to content

docs(contract): lock eval-spec identity - #84

Draft
jeremylongshore wants to merge 2 commits into
mainfrom
feat/iec-E10-eval-substrate-contracts
Draft

docs(contract): lock eval-spec identity#84
jeremylongshore wants to merge 2 commits into
mainfrom
feat/iec-E10-eval-substrate-contracts

Conversation

@jeremylongshore

Copy link
Copy Markdown
Owner

What changed

  • records the canonical EvalSpec identity and the planned SkillEvalSpec profile boundary
  • states that J-Rig's skill-shaped document must be adapted before shared evidence or rollout use
  • adds the repo-local contract identity application document and indexes it
  • corrects the kernel README and CLAUDE version references from 0.9.0 to the shipped 0.10.0
  • records the documentation change in the Unreleased changelog

Why

The kernel is already at @intentsolutions/core@0.10.0, but its docs still reported 0.9.0. More importantly, J-Rig currently calls its skill-specific schema EvalSpec, while the kernel exports a different canonical EvalSpec. This PR makes the boundary explicit before the adapter work lands, preventing a second unrelated source of truth.

Tracking

  • Plan: IEP-EVAL-EVOLUTION-001
  • Refs bead: bd_000-projects-htjt.1
  • Parent repo epic: bd_000-projects-htjt.10
  • Branch: feat/iec-E10-eval-substrate-contracts
  • Commit: de41909

Validation

  • targeted Prettier check on changed files
  • pnpm run check:version-lockstep
  • pnpm run lint
  • pnpm run typecheck
  • pnpm run test — 49 files, 1,412 passed, 1 skipped
  • pnpm run build
  • pre-commit escape-scan and boundaries passed

The repo-wide pnpm run check reaches format:check but is currently blocked by pre-existing ignored python/.mypy_cache files being included in the repository-wide Prettier sweep; no changed source file is implicated.

Records the kernel-versus-skill-profile boundary and updates the repo's published version references to 0.10.0.

Plan: IEP-EVAL-EVOLUTION-001
Bead: bd_000-projects-htjt.1
Repo: intent-eval-core
Depends-on: bd_000-projects-htjt.9
Evidence: targeted prettier check; check:version-lockstep; lint; typecheck; test; build
Known: full check reaches format:check but existing ignored python/.mypy_cache files fail the repo-wide Prettier sweep
@jeremylongshore
jeremylongshore force-pushed the feat/iec-E10-eval-substrate-contracts branch from de41909 to 068d4cd Compare August 2, 2026 01:42

Copy link
Copy Markdown
Owner Author

Maintenance update: rebased feat/iec-E10-eval-substrate-contracts onto current origin/main and force-with-lease pushed. New head: 068d4cd (the contract/docs diff is unchanged). Existing validation evidence remains: version lockstep, lint, typecheck, tests (49 files / 1,412 passed, 1 skipped), build, boundary checks, and escape-scan passed; repo-wide format check remains subject to the documented pre-existing ignored python/.mypy_cache issue.

Plan: IEP-EVAL-EVOLUTION-001
Bead: bd_000-projects-htjt.1
GitHub: intent-eval-core#84

Treat the established platform term rollout as valid prose vocabulary so the new contract-identity document does not create an advisory Vale failure.

Copy link
Copy Markdown
Owner Author

Follow-up pushed as 084c6bc: added the established platform term rollout to the Vale acceptance vocabulary. This removes the diff-specific advisory Vale findings from the new contract-identity document; the commit hook boundary check passed and CI has been re-triggered.

@jeremylongshore

Copy link
Copy Markdown
Owner Author

Verification update: PR #84 is clean with all reported checks green after 084c6bc.

  • Contract identity is explicit: core EvalSpec remains canonical; J-Rig uses SkillEvalSpec.
  • Current package and Python distribution are version-locked at 0.10.0.
  • CI version-lockstep, authoring/codegen parity, boundary, prose-anchor, lint, typecheck, test, build, and CodeQL lanes are green.
  • The dependent J-Rig contract branch has now also removed its diff-specific Vale spelling finding in commit 45160cc.

This satisfies the core-side acceptance for bead bd_000-projects-htjt.1; the PR remains draft until the coordinated stack is ready.

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.

1 participant