CANP-WP-0002 T04: eval schema, render checks and output criteria

`evals/` was a reserved path holding unvalidated blobs: section 12 named the
directory and gave an illustrative snippet, but nothing was specified, so no
tool could act on an eval file.

Every eval file must now declare a `schema`, and CPF defines exactly one —
`canned-prompts/eval-rubric/v0.1`. Unrecognized schemas stay legal and are
skipped rather than rejected, so the format gains something actionable without
becoming an evaluation language, which remains a non-goal.

The schema splits along the same seam as T01 and T03. Render checks
(`contains`, `not_contains`, `resolves_all`) assert properties of the rendered
prompt text, need no model, and are therefore run by the reference CLI. Output
criteria describe a good result and are declared but not run, because judging
them requires a model. That division is now the format's consistent answer to
"deterministic locally, or not".

An eval references a fixture already declared in the manifest's `examples`
rather than carrying its own copy, so an example that is also an eval fixture
stays honest — both break together.

An eval declares assessment and must not record outcomes. Results are run
evidence and live outside the immutable package, per INTENT.md and section 17.

Spec: 12 rewritten with 12.1, 18 (rules 17-18), 21 (`eval` verb).

Reference CLI: read_eval, validate_eval, load_example_values,
run_render_checks, cmd_eval; a failed render check exits non-zero.
Tests 42 -> 51.

examples/pqrst-estimate/evals/quality.yaml is a real eval with four render
checks and four output criteria, and it passes.

Fixes a latent bug reaching a fixture exposed: coerce_value assumed every
value was a command-line string, so a YAML fixture carrying a real type
(include_rationale: true) crashed on .lower(). Typed values are now validated
but not re-parsed.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Bjefh8NUiEiahN4JLwoSKM

Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 388925@bnt-lap001
Assistant-Session: 3507023f-e0fd-4a1e-9d90-a0d4217d1502

README.md

@@ -114,6 +114,32 @@ There is no template inheritance. A package never extends another or overrides
 its parts; composition is by reference only, so a package's content stays
 readable without chasing ancestors.
 
+## Evals
+
+An eval file declares what to assess. `evals/` used to hold whatever an author
+put there; it now has one schema the tooling understands, split along the same
+line as composition:
+
+- **render checks** — deterministic assertions about the *rendered prompt*
+  (`contains`, `not_contains`, `resolves_all`). No model needed, so the
+  reference CLI runs them.
+- **output criteria** — statements about a good *result*. Declared, not run.
+
+```bash
+python canned_prompts.py eval practice/pqrst-estimate
+```
+
+```text
+local:practice/pqrst-estimate@0.2.0
+  evals/quality.yaml (pqrst-estimate-quality)
+    render  PASS  contains "must sum to exactly 100%"
+    render  PASS  resolves_all
+    output  --    4 criteria declared (not run: judging output needs a model)
+```
+
+An eval declares assessment, never results. Results are run evidence and live
+outside the immutable package.
+
 ## Registries and identity
 
 An id names a package *within a registry* (`CannedPromptFormat-v0.1.md`