Skip to content

PR7: boulder-native-preview profile and docs integration - #35

Open
min9lin9 wants to merge 51 commits into
boulder-native-planner/pr6-execution-packetfrom
boulder-native-planner/pr7-preview-integration
Open

PR7: boulder-native-preview profile and docs integration#35
min9lin9 wants to merge 51 commits into
boulder-native-planner/pr6-execution-packetfrom
boulder-native-planner/pr7-preview-integration

Conversation

@min9lin9

Copy link
Copy Markdown
Owner

Summary

  • Explicit boulder-native-preview profile with doctor and quickstart visibility; preview event evidence local-only
  • Package allowlist coverage and README/architecture/doctor documentation
  • programming-default byte-equivalent; Handoff v1 unchanged; external Handoff bridge remains follow-up RFC

Merge gate

  • bun test: 361 pass / 3116 assertions
  • bunx tsc --noEmit

Stacked on PR6.

Burt and others added 30 commits July 18, 2026 16:26
Adds the explicit boulder-native-preview profile with doctor and
quickstart visibility, preview event evidence kept local-only, package
allowlist coverage, and README/architecture/doctor documentation.
programming-default resolution stays byte-equivalent and Handoff v1 is
unchanged; the external Handoff bridge remains follow-up RFC work.

Merge gate: bun test (361 pass), bunx tsc --noEmit.
Adds plan benchmark with explicit trust-root/study-root inputs, Ed25519
provenance verification, descriptor-pinned no-follow evidence reads,
canonical study identity and report recomputation, the strict common
planner-output normalizer with independently verified source trust, and
deterministic HOLD-first promotion decisions. Adds the plan benchmark
subcommand, fixtures, and benchmark plan documentation. The repository
never contacts providers or executes study runs; PR8B operational
evidence stays external, and the recorded v0.2 field-study outcome is
HOLD (user action request included).

Merge gate: bun test (403 pass), bunx tsc --noEmit, 222-file pack.
…ulder-native-planner/pr8a-benchmark-tooling
Self-contained src/v2 kernel (JCS digests, effect gating, authority
verifier, lifecycle) with boulder v2 execute CLI, frozen v2-kernel
fixtures, and authority-vector corpus. ADR 0003 clause on root
AGENTS.md is scoped from byte-freeze to gate-outcome claims, with
counterevidence recorded (committed planner-stack docs predated it).

Refs: docs/adr/0003-v2-kernel-gates.md (K1 bounded; K2-K4 separate gates)
Byte-first reader (64 KiB limit, BOM/UTF-8/duplicate-key rejection
before JSON.parse), canonical digests, and exact-issue validation for
the frozen k2a-f contract foundation (fixtures/k2a-f v1).
… remediation, executor evidence

Adds the pre-execution safety gate (signed receipts), scope-attribution
receipts, score workflow state machine (blinding/lock/reveal), study
remediation policy wired additively into planner-benchmark, and the
common executor lifecycle/receipt evidence schema. Optional contract
fields only; existing bundles unchanged.
Isolated sandbox runner, independent byte-level oracle, and capture
pipeline producing the k0r evidence manifest. Evidence artifacts under
evidence/k0r/ are regenerated against the settled tree in a follow-up
commit.
Root and subdir guides updated for the planner stack, v2 kernel, k2a-f,
k0r harness, and ops router; new scoped guides for fixtures/ and
evidence/. Documents the tree as it exists per amended ADR 0003 clause
(no gate-outcome claims).
Adds reference/Boulder_Reference_Implementation_Strategy v0.2 (read-only
reference) and README --dry-run corrections. Moves the ReFoundation
planning zip from docs/ to plans/ so it stays out of the npm package
surface; doc-registry entry removed. The two Korean session/gap docs
stay untracked: they are gitignored local-only notes by design.
… tree

Reclassifies packed files after the v2/k2a-f/planner landings (runtime
107, fixture 44, public-doc 66; totals 249/250), drops the quarantined
ReFoundation zip from the packaged set, classifies the three new
AGENTS.md docs consistent with existing shipped AGENTS.md files, and
refreshes the readiness-v0 gate baselines from live evaluations.
Q3 conditions met: Q2 dispositions applied (8 commits), v1 deltas classified additive/expected, ADR 0003 clause amended with counterevidence, baselines regenerated, suite green except seven k0r evidence-contract tests whose recapture is gated on the ADR 0004 human exit flow (evidence/k0r/ held untracked as needs-evidence). 10732cb remains the historical AS-IS reference; Boulder-9-3-plus stays the read-only comparison control for that revision.
Traces every strategy section 9 fitness function to an enforced test
suite or a namespaced planned REF-E/REF-PR experiment so prose cannot
silently become a claim of enforcement. New executable guards: no
runtime-host package imports anywhere in src/, k2a-f sibling-only .js
imports, plan/planner cohort explicit-.js specifiers, no process.exit,
zero runtime dependencies, v2 K2-K4 anti-overclaim, and doctor probe
non-mutation. Also fixes the one pre-existing cohort violation the new
guard caught (plan-receipts type import missing .js).
- Replace boulder guide with ko html version bound to evidence harness
- Add k0r baseline generator, canonical helpers, issue-exit, reconcile
- Regenerate package inventory and readiness baselines for landed tree
- Make namespace-hostile tests pass under production bwrap CI: route
  package dry-run via writable staging, guard fresh-user K0R contract
  tests behind BOULDER_CI_BWRAP, adapt nested-bwrap probe
…ue path

Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
…guard

Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
min9lin9 and others added 20 commits August 26, 2026 10:47
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
…hmark-tooling

PR8A: Signed planner benchmark evidence tooling
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
The fresh K0R authority (scope 87d3019 lineage, approved 2026-08-28) requires the
prior exit receipt and final-verification bundle to be absent from the index and
worktree: the capture contract's artifact discovery and dirty assessment admit no
tracked k0r-namespace path outside the approved overlay set. Both bytes are
preserved byte-identical under the private QA authority root for rollback.
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
The fresh K0R authority (scope 87d3019 lineage, approved 2026-08-28) requires the
prior exit receipt and final-verification bundle to be absent from the index and
worktree: the capture contract's artifact discovery and dirty assessment admit no
tracked k0r-namespace path outside the approved overlay set. Both bytes are
preserved byte-identical under the private QA authority root for rollback.
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
Ultraworked with [omo](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: sisyphus-dev-ai <sisyphus-dev-ai@users.noreply.github.com>
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