feat(ai): flip tool-argument validation to native

[zmaril]

What

Flips pi's packages/ai/src/utils/validation.ts (tool-argument validation +
JSON-Schema coercion) to the native Rust engine. validateToolArguments and
validateToolCall now run in crates/pidgin-ai (utils/validation.rs);
packages/ai/test/validation.test.ts passes fully native.

Before / After

Before: validation.ts ran pi's own TypeScript — TypeBox Value.Convert +
the hand-rolled coerceWithJsonSchema + compiled Value.Check. Manifest row
status: "original". Native module count: 52.

After: a native shim overlays the module and delegates to the addon. Every
decision runs in Rust: the AJV-compatible primitive-coercion rules
(coercePrimitiveByType), the recursive allOf/anyOf/oneOf + object + array
coercion driver, the pass/fail check over the JSON-Schema subset, and the
Validation failed for tool "<name>": … throw-envelope. Manifest row
status: "native", tests: ["test/validation.test.ts"]. Native module count: 53.

How

• crates/pidgin-napi/src/ai_validation.rs (new): exposes both functions as
  free #[napi] fns. Tool / ToolCall cross the boundary as JSON strings —
  the tool's parameters is a TypeBox TSchema, and JSON.stringify drops its
  hidden [TypeBox.Kind] symbol, leaving exactly the JSON-Schema projection the
  port models as serde_json::Value. Each fn returns a discriminated outcome
  envelope: {ok:true,value} on success, {ok:false,error} carrying pi's thrown
  message verbatim.
• crates/pidgin-napi/src/lib.rs: registers pub mod ai_validation; (pub
  for the free-fn crate-reachability rule).
• conformance/shims/packages/ai/src/utils/validation.ts (new): re-exports
  the preserved pi original, overrides both fns to JSON.stringify their inputs,
  delegate, return the coerced value, and re-throw new Error(envelope) on
  failure so pi's .toThrow("Validation failed") / Tool "<name>" not found
  contract holds. .toEqual deep-equality means the freshly parsed return is fine.
• conformance/manifest.json: the one validation.ts row flipped to native.

The symbol-drop is behaviour-preserving

pi's coercion path has two branches gated on [TypeBox.Kind]: TypeBox schemas
run only Value.Convert, plain serialized schemas additionally run the
hand-rolled coercion. Rust Values carry no Kind, so the port always runs the
hand-rolled coercion — and for the shapes these utils see it reproduces the same
observable result Value.Convert would have produced (e.g. {count:"42"} ->
{count:42} for a Type.Number() field). The tests assert only plain coerced
values (.toEqual) and, on failure, only the substring "Validation failed"
(pi's own envelope prefix) — never TypeBox's per-error message text — so the
reimplementation satisfies the full test contract.

Verification

• test/validation.test.ts: 3/3 pass, native (all cases).
• Full packages/ai suite: zero new regressions — 553 passed / 3 failed /
  740 skipped; the 3 failures are pre-existing model-catalog phantoms
  (Anthropic adaptive-thinking + Kimi pricing, from generated model data),
  unrelated to this change.
• Codegen missing: 0, native 52 -> 53.
• cargo fmt --all --check, cargo clippy --all-targets (no warnings), and
  whole-repo straitjacket v0.2.3 (crates conformance scripts .github, vendor
  excluded): 0 errors.

🤖 Generated with Claude Code

---

Generated by Claude Code

[github-actions]

Conformance smoke: agent + ai + tui

• pi_sha: 3da591ab74ab9ab407e72ed882600b2c851fae21
• manifest_native_modules: 55

Headline is rust-backed: passing cases in files whose module-under-test is a native (Rust addon) module. Raw all-pass is shown secondary — it is inflated by unflipped TypeScript that passes without touching any Rust.

Package	rust-backed	Native modules	raw pass (secondary)	Failing	Skipped
agent	73/180 (40.6%)	7	180/180	0	0
ai	84/1296 (6.5%)	12	553/1296	3	740
tui	375/678 (55.3%)	14	678/678	0	0
smoke total	532/2154 (24.7%)	33	1411/2154	3	740

agent: 73/180 (40.6%) · ai: 84/1296 (6.5%) · tui: 375/678 (55.3%)

rust-backed = passing / total tests run, per the manifest's per-native-row tests lists. The Native modules column counts modules served by the Rust addon. This is the agent+ai+tui smoke subset; the full baseline lives in committed conformance.json.

Attribution

A pi test file is rust-backed iff the module it primarily exercises — its module-under-test — is status=native in conformance/manifest.json. A package's rust-backed count is the passing cases in its rust-backed files. Transitive or infrastructure use does not count (e.g. ~30 ai files construct the native faux provider but test other things — only faux-provider.test.ts, whose subject is faux, counts). A file that substantially tests both a native and an original module is excluded rather than counted, so the number under-reports rather than over-claims.

Per-file decisions (from each native manifest row's tests list):

Native module	Test file	Decision
ai/api/anthropic-messages.ts	test/anthropic-sse-parsing.test.ts	counted
ai/providers/faux.ts	test/faux-provider.test.ts	counted
coding-agent/utils/ansi.ts	test/ansi-utils.test.ts	counted
coding-agent/utils/changelog.ts	test/changelog.test.ts	counted
coding-agent/utils/git.ts	test/git-ssh-url.test.ts	counted
coding-agent/core/tools/path-utils.ts	test/path-utils.test.ts	counted
tui/keys.ts	test/keys.test.ts	counted
tui/utils.ts	test/truncate-to-width.test.ts, test/regression-regional-indicator-width.test.ts	counted
coding-agent/utils/mime.ts	test/image-process.test.ts	mixed — excluded (2 of 3 cases test processImage/original)
coding-agent/utils/version-check.ts	test/version-check.test.ts	mixed — excluded (fetch cases mock the original; not separable per-file)
coding-agent/core/export-html/ansi-to-html.ts	test/export-html-whitespace.test.ts	mixed — excluded (asset-grep + tool-renderer/original dominate)
coding-agent/core/tools/truncate.ts	—	no dedicated test (exercised via tools.test.ts, subject is the read/edit factory/original)
coding-agent/core/tools/edit-diff.ts	—	no dedicated test (edit-tool tests' subject is the edit tool/original)

CLI conformance (black-box, against the pidgin binary)

• CLI conformance: 15/15 pass against target/release/pidgin (pass delta ±0, fail delta ±0).

File	Passing	Failing	Skipped
packages/coding-agent/test/session-file-invalid.test.ts	1	0	0
packages/coding-agent/test/session-id-readonly.test.ts	7	0	0
packages/coding-agent/test/startup-session-name.test.ts	2	0	0
packages/coding-agent/test/stdout-cleanliness.test.ts	5	0	0

The four repointed coding-agent CLI test files spawn the compiled pidgin binary via $PIDGIN_BIN instead of pi's own cli.ts. This is a separate signal from the module smoke table and is never folded into the per-package Native count.