Bind external requests to canonical schema resources by flyingrobots · Pull Request #181 · flyingrobots/edict

flyingrobots · 2026-07-30T12:03:53Z

Plain-English Walkthrough

TL;DR

External-action source no longer pins placeholder schema or reconciliation digests. [claim:real-request-resources, confidence:1.00] The owning generator now emits canonical input-schema, settlement-schema, and reconciliation-law artifacts for both workspace observation and validated patch requests; the public application build independently validates their shape, recomputes their domain-framed identities, and requires the configured set to equal the compiler-emitted Target request closure.

Invalid resource closures fail before Core or Target IR publication. [claim:fail-closed-resource-closure, confidence:1.00] Missing, duplicate, disconnected, substituted, opaque, non-canonical, sentinel, fixed-seed mutated, and over-budget resource sets have executable rejection witnesses.

Walkthrough

Previously, Core and Target IR faithfully preserved schema and reconciliation coordinates, but the workspace fixtures bound those coordinates to repeated-byte placeholder digests. A caller could therefore present compiler-consistent request artifacts without presenting the real resources those identities claimed to name.

edict.application/v1 now accepts externalActionResources, a bounded list of canonical resource artifact paths. Each artifact declares its coordinate, resource kind, and closed definition under edict.external-action-resource/v1. The build decodes canonical CBOR, validates the schema or reconciliation meta-contract, recomputes the exact domain-framed SHA-256 identity, and compares it to the request slot that cited it. [claim:independent-resource-admission, confidence:1.00]

The authority and data flow is:

flowchart LR
    S[Edict source request] --> C[Compiler Core and Target IR]
    G[Generator-owned resource artifacts] --> A[Public build admission]
    C --> A
    A -->|exact complete closure| P[Publish Core and Target IR]
    A -->|missing or mismatched| R[Typed refusal]

Caption: External request resource admission

Source pins each schema and reconciliation resource identity.
The owning generator emits the canonical resource bytes and regenerates Core and Target IR.
The public build independently derives each resource identity and compares the supplied set with the Target request closure.
Publication is reachable only from an exact complete closure; invalid closure states terminate in a refusal.

This keeps request description separate from performance authority. The new artifacts contain declarative schemas and reconciliation bindings only. No provider component is invoked, no filesystem capability enters Edict semantics, and no adapter operation is performed. [claim:no-performance-authority, confidence:1.00]

Both workspace fixture families are regenerated through cargo xtask lawpack-goldens --write, so the checked source, resource sidecars, Core, and Target IR remain one generator-owned closure. [claim:generator-owned-closure, confidence:1.00] No dependency was added.

Compatibility impact is explicit: external-action application manifests must list every request resource in externalActionResources, and consumers must repin the regenerated Core/Target artifacts. Executable-operation manifests reject a non-empty external resource list. The field is capped at 192 artifacts, matching three distinct resources for a 64-request closure.

RED / GREEN

RED:

cargo test -p edict-cli --bin edict public_external_action_build_rejects_sentinel_resource_identities -- --nocapture

The test failed because build_application returned success for repeated-byte request resource identities.

GREEN:

cargo xtask verify

The full repository gate passed at 663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13: formatting, strict all-feature clippy, workspace and doc tests, every owned golden check, provider fixtures and contract packs, provider dependency boundary, topic contracts, and diff checks.

Appendix: Citations

Claim	Evidence	Confidence	Notes
`claim:real-request-resources`	`xtask/src/lawpack_goldens.rs#1416@663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13`; `xtask/src/lawpack_goldens.rs#1490@663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13`; `xtask/src/lawpack_goldens.rs#1551@663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13`	1.00	Generator constructs canonical typed resources for both request families.
`claim:fail-closed-resource-closure`	`crates/edict-cli/src/application_build.rs#849@663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13`; `public_external_action_build_rejects_sentinel_resource_identities` at `crates/edict-cli/src/application_build.rs#2600@663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13`; `fixed_seed_request_resource_mutations_fail_closed` at `crates/edict-cli/src/application_build.rs#2800@663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13`	1.00	Exact, sentinel, property, and closure refusal paths are executable.
`claim:independent-resource-admission`	`crates/edict-cli/src/application_build.rs#482@663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13`; `crates/edict-cli/src/application_build.rs#535@663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13`; `crates/edict-cli/src/application_build.rs#799@663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13`	1.00	Loading, meta-validation, and request-closure derivation are separate checks.
`claim:no-performance-authority`	`docs/topics/external-action-requests/README.md#69@663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13`; `docs/topics/external-action-requests/README.md#102@663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13`	1.00	Durable topic contract states the request-only authority boundary.
`claim:generator-owned-closure`	`xtask/src/lawpack_goldens.rs#177@663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13`; `xtask/src/lawpack_goldens.rs#472@663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13`; `lawpack_goldens_match_executable_codec` plus `cargo xtask verify` at `663148a1ba1afbdbdf2ea9a7ba6da40d2e9a5d13`	1.00	Owning generator and check mode reproduce all checked bytes.

Closes #180.

chatgpt-codex-connector · 2026-07-30T12:03:57Z

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, add credits to your account and enable them for code reviews in your settings.

flyingrobots · 2026-07-30T12:04:00Z

@coderabbitai review

flyingrobots · 2026-07-30T12:04:02Z

@codex please review

coderabbitai · 2026-07-30T12:04:07Z

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

coderabbitai · 2026-07-30T12:04:10Z

Warning

Review limit reached

@flyingrobots, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 34 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: f67b318a-ce98-4b35-87ba-fc35c33db836

📥 Commits

Reviewing files that changed from the base of the PR and between 663148a and 3148f0a.

📒 Files selected for processing (9)

CHANGELOG.md
crates/edict-cli/src/application_build.rs
crates/edict-syntax/src/canonical.rs
crates/edict-syntax/src/lib.rs
docs/topics/cli/README.md
docs/topics/cli/test-plan.md
docs/topics/external-action-requests/test-plan.md
fixtures/lawpack/workspace-patch/README.md
xtask/src/lawpack_goldens.rs

Summary by CodeRabbit

New Features
- External-action builds now validate canonical, digest-locked resource artifacts before publication.
- Added support for input schemas, settlement schemas, and reconciliation laws through explicit resource listings.
- Added a maximum limit of 192 external-action resources.
Bug Fixes
- Builds now reject missing, duplicate, disconnected, substituted, opaque, malformed, placeholder, or mutated resources.
Documentation
- Updated CLI, external-action, fixture, and test-plan documentation with the strengthened validation contract.

Walkthrough

External-action builds now require generator-owned canonical CBOR resource artifacts. The CLI loads and independently validates bounded externalActionResources closures, matches them to digest-locked requests, and updates Lawpack fixtures, golden outputs, documentation, and refusal tests.

Changes

External-action resource validation

Layer / File(s)	Summary
Canonical resource generation `xtask/src/lawpack_goldens.rs`	Generates input schemas, settlement schemas, reconciliation laws, domain-separated digests, and updated Core/Target IR goldens for workspace snapshot and patch applications.
Build loading and closure validation `crates/edict-cli/src/application_build.rs`	Loads canonical resource artifacts from the manifest, validates their structures and digests, enforces the 192-resource limit, and requires an exact request-referenced closure.
Validation fixtures and tests `crates/edict-cli/src/application_build.rs`, `fixtures/lawpack/...`	Adds resource fixtures and tests for valid builds, placeholder or mutated identities, opaque and non-canonical definitions, missing or disconnected resources, duplicates, and boundary overflow.
CLI and external-action contracts `CHANGELOG.md`, `docs/topics/cli/`, `docs/topics/external-action-requests/`, `fixtures/lawpack/*/README.md`	Documents canonical resource resolution, rejection conditions, provider non-invocation, atomic publication, and updated digest-locked request examples.

Estimated code review effort: 4 (Complex) | ~45 minutes

Possibly related issues

flyingrobots/edict#180 — Covers generator-owned workspace-patch schema artifacts, sentinel replacement, and build-time resource substitution rejection.

Possibly related PRs

flyingrobots/edict#179 — Directly overlaps the Lawpack resource generation and external-action schema/digest validation flow.
flyingrobots/edict#177 — Extends the same request-only external-action build and canonical publication pipeline.
flyingrobots/edict#170 — Provides the related verified build_application provider lowerer/verifier flow.

Poem

Three schemas march in canonical light,
Their digests guard the build outright.
No ghosts, no gaps, no borrowed claim—
The closure locks each artifact’s name.
Core and Target rise only when true.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title is concise and accurately summarizes the main change: binding external requests to canonical schema resources.
Description check	✅ Passed	The description is directly related to the changeset and explains the new canonical external-action resource flow.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands.}

chatgpt-codex-connector · 2026-07-30T12:04:10Z

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, add credits to your account and enable them for code reviews in your settings.

coderabbitai

Actionable comments posted: 8

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@crates/edict-cli/src/application_build.rs`:
- Around line 2799-2848: Replace
fixed_seed_request_resource_mutations_fail_closed with a single representative
mutation, removing RESOURCE_MUTATION_SEED and the ordinal loop, since every
iteration currently changes only definition.root and asserts the same outcome.
Keep the existing temp-tree setup, artifact mutation, build_application call,
and ExternalActionResourceClosureMismatch assertion for that one case.
- Around line 43-44: Centralize EXTERNAL_ACTION_RESOURCE_API_VERSION and
EXTERNAL_ACTION_RESOURCE_DIGEST_DOMAIN as pub const values in
edict-syntax/src/canonical.rs beside the existing digest-domain constants, and
re-export them from the edict_syntax crate root. In
crates/edict-cli/src/application_build.rs lines 43-44 and
xtask/src/lawpack_goldens.rs lines 15-17, remove the local duplicate constants
and import both shared symbols from edict_syntax; leave DIGEST_FRAME unchanged
in xtask.
- Around line 519-521: Replace the unreachable! branch in the build path around
the definition field lookup with the existing ApplicationBuildFailure
constructor, returning a structured failure with the appropriate stable error
kind when "definition" is absent. Preserve the successful field extraction and
avoid relying on the fields.get call being guaranteed by the string-literal
field set.
- Around line 2785-2792: Update the refusal-case table and its assertion in the
relevant test so each case carries its expected failure kind: missing,
duplicate, and disconnected must require ExternalActionResourceClosureMismatch,
while noncanonical must require InvalidExternalActionResource. Replace the broad
matches! acceptance with an equality check against the case-specific
expectation, preserving the existing failure reporting.

In `@docs/topics/external-action-requests/test-plan.md`:
- Around line 82-86: Standardize this rejection class on “sentinel” terminology,
preserving EXTREQ-TP-026 in
docs/topics/external-action-requests/test-plan.md:82-86 as the canonical
wording. Update the corresponding “placeholder identities” wording in
docs/topics/cli/test-plan.md:106-107, docs/topics/cli/README.md:97-107, and
CHANGELOG.md:13-18 to use “sentinel” consistently, including the related
resource descriptions.

In `@fixtures/lawpack/workspace-patch/README.md`:
- Around line 14-20: Add [claim:<id>, confidence:<value>] tags to the first
occurrences of the technical claims in the README section, and append the
required collapsed citations appendix. Cite xtask/src/lawpack_goldens.rs:478-583
and the repository-relative loader implementation that validates the complete
closure, including line ranges and Git SHAs.

In `@xtask/src/lawpack_goldens.rs`:
- Around line 433-438: Update xtask/src/lawpack_goldens.rs:433-438
workspace_snapshot_application_source and xtask/src/lawpack_goldens.rs:731-736
workspace_patch_application_source to accept GeneratedExternalActionResource
references instead of loose digest strings; add coordinate: String to
GeneratedExternalActionResource at xtask/src/lawpack_goldens.rs:185-199 and
populate it from external_action_resource; update callers at
xtask/src/lawpack_goldens.rs:226-233 to pass resource values, and make both
templates use each resource’s coordinate and digest rather than duplicated
hardcoded coordinates (snapshot sites 461, 463, 470; patch sites 759, 761, 768).
- Around line 1452-1458: Update the non-Lowered branch after lower_to_target_ir
in the lawpack golden generation flow to include the returned failures
diagnostics in the error message, following the existing build_application
pattern. Preserve the status reporting while formatting the failures payload so
callers receive the underlying causes instead of only TargetLoweringStatus.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

Push a commit to this branch (recommended)
Create a new PR with the fixes

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 8157d51d-ebf1-4e15-9322-913261e048b0

📥 Commits

Reviewing files that changed from the base of the PR and between cf8c17f and 663148a.

📒 Files selected for processing (31)

CHANGELOG.md
crates/edict-cli/src/application_build.rs
docs/topics/cli/README.md
docs/topics/cli/test-plan.md
docs/topics/external-action-requests/README.md
docs/topics/external-action-requests/test-plan.md
fixtures/lawpack/workspace-patch/README.md
fixtures/lawpack/workspace-patch/apply-validated-patch.core.cbor
fixtures/lawpack/workspace-patch/apply-validated-patch.core.sha256
fixtures/lawpack/workspace-patch/apply-validated-patch.edict
fixtures/lawpack/workspace-patch/apply-validated-patch.target-ir.cbor
fixtures/lawpack/workspace-patch/apply-validated-patch.target-ir.sha256
fixtures/lawpack/workspace-patch/input-schema.cbor
fixtures/lawpack/workspace-patch/input-schema.sha256
fixtures/lawpack/workspace-patch/reconciliation-law.cbor
fixtures/lawpack/workspace-patch/reconciliation-law.sha256
fixtures/lawpack/workspace-patch/settlement-schema.cbor
fixtures/lawpack/workspace-patch/settlement-schema.sha256
fixtures/lawpack/workspace-snapshot/README.md
fixtures/lawpack/workspace-snapshot/input-schema.cbor
fixtures/lawpack/workspace-snapshot/input-schema.sha256
fixtures/lawpack/workspace-snapshot/observe-workspace.core.cbor
fixtures/lawpack/workspace-snapshot/observe-workspace.core.sha256
fixtures/lawpack/workspace-snapshot/observe-workspace.edict
fixtures/lawpack/workspace-snapshot/observe-workspace.target-ir.cbor
fixtures/lawpack/workspace-snapshot/observe-workspace.target-ir.sha256
fixtures/lawpack/workspace-snapshot/reconciliation-law.cbor
fixtures/lawpack/workspace-snapshot/reconciliation-law.sha256
fixtures/lawpack/workspace-snapshot/settlement-schema.cbor
fixtures/lawpack/workspace-snapshot/settlement-schema.sha256
xtask/src/lawpack_goldens.rs

📜 Review details

⏰ Context from checks skipped due to timeout. (3)

GitHub Check: supply-chain (cargo-deny)
GitHub Check: rust stable (fmt · clippy · test)
GitHub Check: rust msrv 1.94.0 (fmt · clippy · test)

🧰 Additional context used

📓 Path-based instructions (6)

**/*