# Film and surface rule evaluator

## What this pack lets workers do

A collection of nouns and verbs for evaluating declarative film and surface rule nouns (`<`, `<=`, `>=`, `abs<=`, `in_range`, `empty`, `is_true`, `is_false` with honest `unknown` propagation), blocking forbidden synthetic or non-functional stamps at native frame resolution, scoring frosted-pill caption centering and dual-channel competing cognitive load, repointing dead external repository links to SHA-256-pinned `/source/` mirror routes, and verifying `/llms.txt` tool-count parity against live MCP `tools/list`.

## Nouns

| Noun (task object) | What it holds |
| --- | --- |
| Guardrails | Forbid fabricated zero values when measurements are absent (`unknown`); forbid synthetic/non-functional stamps or simulated fallback answers; require `distinct_reviewer` before admitting a render or surface cut |
| `spec.declarative-film-and-surface-rules` | Eight declarative comparison operators, `pre`/`post` hook stages, frosted-pill geometry bounds and dual-channel reading-rate limits |
| `rules.mirror-relink-and-mcp-parity` | Deterministic `/source/` mirror anchor/archive/intake mapping rules and `/llms.txt` vs MCP `tools/list` 1:1 parity rules |
| `tool.film-surface-rules` | Tested standard-library Python package in [code/](code/README.md), pinned by path, file count and digest |
| `template.film-surface-rules-receipt` | JSON Schema for rule verdicts, stamp/chip scan results, caption/cognitive-load checks, mirror rewrites and MCP parity |
| Worked example | Synthetic cut and surface bundle (`example.film-surface-rules-bundle`) and recomputed receipt (`example.film-surface-rules-receipt`) |

## Verbs

| Verb (task) | Who | What happens |
| --- | --- | --- |
| `evaluate-film-rules-and-caption-load` | agent, no model | Evaluate declarative rule nouns, scan native-resolution OCR spans for forbidden stamps and score frosted-pill centering and dual-channel load |
| `repoint-dead-links-and-check-mcp-parity` | agent, no model | Rewrite dead repository URLs to SHA-256-pinned `/source/` mirror routes and verify `/llms.txt` parity against MCP `tools/list` |
| `review-film-and-surface-readiness` | agent, no model | Independently verify zero block failures, zero simulated fallbacks and complete mirror coverage (`distinct_reviewer`) |

## How to use it

Run `make check` in [code/](code/README.md) to run all unit tests covering all eight rule operators, `unknown` measurement honesty, forbidden-stamp detection, dual-channel load collisions, `/source/` mirror link rewriting, MCP parity, `ZeroHardcodedValuesAstGate` AST scanning and worked-example receipt recomputation.

## Limits

- Quarantined, unverified proposal until reviewed: this pack is not installed, imported, admitted or activated, and no worker has claimed or run these tasks through Tessera.
- Offline rule evaluation, link rewriting and parity checks run deterministically without network access or live service mutation.
- Threshold pass verdicts verify measured geometric, loudness and parity constraints; people still watch cuts and inspect pages for editorial clarity.