frontface-proofit 1.2.0__tar.gz → 1.2.1__tar.gz

{frontface_proofit-1.2.0/src/frontface_proofit.egg-info → frontface_proofit-1.2.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: frontface-proofit
-Version: 1.2.0
+Version: 1.2.1
 Summary: Fail-closed linter for locked-down runbooks: ten rules of workflow discipline as code
 Author: Ryan Wesley Hitchcock
 License: MIT
@@ -21,8 +21,10 @@ people following it do the same thing and get the same result. proofit checks
 a runbook draft against fourteen rules of workflow discipline — ten structural
 rules plus the 1337 module's four (cross-role duplication, header date,
 suspicious character count, missing-file verification) — and **rejects the
-draft if any rule is violated**. No warnings, no advisories. Fail-closed:
-exit 1 until it's fixed, exit 0 when it's locked down.
+draft if any structural rule is violated**. Heuristic and convention rules
+(R1, R5, R8, R13) come back as warnings — shown, but they don't fail the
+draft unless `--strict` is passed. Fail-closed where it counts:
+exit 1 until the structural rules pass, exit 0 when it's locked down.
 
 It checks the runbook's **shape, not its truth**: it can tell you a step is
 missing its verification block, not whether the verification would actually
@@ -74,10 +76,11 @@ mechanical: the rules that can be checked as structure *are* checked as
 structure, so compliance isn't a matter of discipline — it's a matter of
 passing the gate.
 
-Two rules (1 and 8) are phrase-list heuristics — tripwires, not proof. Under
-fail-closed they still block, so they are deliberately narrow: they fire only
-on clear hits. A foolproof gate that misfires on good drafts isn't foolproof,
-it's a tax. The narrowness is documented in the code and tested.
+Two rules (1 and 8) are phrase-list heuristics — tripwires, not proof. They
+report as warnings in default mode and block only under `--strict`, so they
+are deliberately narrow: they fire only on clear hits. A foolproof gate that
+misfires on good drafts isn't foolproof, it's a tax. The narrowness is
+documented in the code and tested.
 
 ## Quickstart
 
@@ -85,14 +88,20 @@ Zero dependencies beyond Python 3.10+.
 
 ```bash
 cd proofit
-PYTHONPATH=src python3 -m unittest discover -s tests   # 65 offline tests
+PYTHONPATH=src python3 -m unittest discover -s tests   # 90 offline tests
 ```
 
-Lint a draft:
+Lint a known-good draft first (expect PASS: no violations):
+
+```bash
+PYTHONPATH=src python3 -m proofit lint examples/fixed-runbook.md
+```
+
+The blank template REJECTS until its placeholders are real observable checks. Then lint a draft:
 
 ```bash
 PYTHONPATH=src python3 -m proofit lint my-runbook.md
-# violations listed as RULE n [location] ERROR — message + fix
+# violations listed as RULE n [location] ERROR|WARNING — message + fix
 # exit 1: REJECTED — fix and re-run
 # exit 0: PASS — the runbook is locked down
 ```
@@ -127,9 +136,10 @@ fixed draft passes).
   than producing meaningful per-rule violations. Write it in the template,
   get it checked — anything else isn't a bad runbook to proofit, it's not
   a runbook at all.
-- **Output:** a list of violations (`RULE n [location] ERROR — message`
-  plus a `fix:` line), and an exit code: 1 if any violation exists, 0 if
-  clean. Fail-closed — there is no warning level.
+- **Output:** a list of findings (`RULE n [location] ERROR|WARNING — message`
+  plus a `fix:` line), and an exit code: 1 if any ERROR exists, 0 if
+  clean. Structural violations are fail-closed; heuristic and convention
+  hits are warnings unless `--strict`.
 
 ## Worked example
 
@@ -155,7 +165,11 @@ linter.
 
 **Why fail-closed instead of warnings?** Because the skill's promise is
 "unbreakable, zero-assumption directive." A warning you can ignore is the
-opposite of idiot-proof. The gate is the product.
+opposite of idiot-proof — so structural violations are fail-closed, no
+exceptions. The heuristic and convention rules (R1, R5, R8, R13) are the
+narrow exception: they report as warnings because they're tripwires, not
+proof, and tripwires can false-positive. Pass `--strict` to make them fail
+the draft too. The gate is still the product.
 
 **Single-service workflow — do I need the routing section?** No. Rule 9 is
 vacuous when no services are named. Delete the Routing Architecture section

{frontface_proofit-1.2.0 → frontface_proofit-1.2.1}/README.md

@@ -5,8 +5,10 @@ people following it do the same thing and get the same result. proofit checks
 a runbook draft against fourteen rules of workflow discipline — ten structural
 rules plus the 1337 module's four (cross-role duplication, header date,
 suspicious character count, missing-file verification) — and **rejects the
-draft if any rule is violated**. No warnings, no advisories. Fail-closed:
-exit 1 until it's fixed, exit 0 when it's locked down.
+draft if any structural rule is violated**. Heuristic and convention rules
+(R1, R5, R8, R13) come back as warnings — shown, but they don't fail the
+draft unless `--strict` is passed. Fail-closed where it counts:
+exit 1 until the structural rules pass, exit 0 when it's locked down.
 
 It checks the runbook's **shape, not its truth**: it can tell you a step is
 missing its verification block, not whether the verification would actually
@@ -58,10 +60,11 @@ mechanical: the rules that can be checked as structure *are* checked as
 structure, so compliance isn't a matter of discipline — it's a matter of
 passing the gate.
 
-Two rules (1 and 8) are phrase-list heuristics — tripwires, not proof. Under
-fail-closed they still block, so they are deliberately narrow: they fire only
-on clear hits. A foolproof gate that misfires on good drafts isn't foolproof,
-it's a tax. The narrowness is documented in the code and tested.
+Two rules (1 and 8) are phrase-list heuristics — tripwires, not proof. They
+report as warnings in default mode and block only under `--strict`, so they
+are deliberately narrow: they fire only on clear hits. A foolproof gate that
+misfires on good drafts isn't foolproof, it's a tax. The narrowness is
+documented in the code and tested.
 
 ## Quickstart
 
@@ -69,14 +72,20 @@ Zero dependencies beyond Python 3.10+.
 
 ```bash
 cd proofit
-PYTHONPATH=src python3 -m unittest discover -s tests   # 65 offline tests
+PYTHONPATH=src python3 -m unittest discover -s tests   # 90 offline tests
 ```
 
-Lint a draft:
+Lint a known-good draft first (expect PASS: no violations):
+
+```bash
+PYTHONPATH=src python3 -m proofit lint examples/fixed-runbook.md
+```
+
+The blank template REJECTS until its placeholders are real observable checks. Then lint a draft:
 
 ```bash
 PYTHONPATH=src python3 -m proofit lint my-runbook.md
-# violations listed as RULE n [location] ERROR — message + fix
+# violations listed as RULE n [location] ERROR|WARNING — message + fix
 # exit 1: REJECTED — fix and re-run
 # exit 0: PASS — the runbook is locked down
 ```
@@ -111,9 +120,10 @@ fixed draft passes).
   than producing meaningful per-rule violations. Write it in the template,
   get it checked — anything else isn't a bad runbook to proofit, it's not
   a runbook at all.
-- **Output:** a list of violations (`RULE n [location] ERROR — message`
-  plus a `fix:` line), and an exit code: 1 if any violation exists, 0 if
-  clean. Fail-closed — there is no warning level.
+- **Output:** a list of findings (`RULE n [location] ERROR|WARNING — message`
+  plus a `fix:` line), and an exit code: 1 if any ERROR exists, 0 if
+  clean. Structural violations are fail-closed; heuristic and convention
+  hits are warnings unless `--strict`.
 
 ## Worked example
 
@@ -139,7 +149,11 @@ linter.
 
 **Why fail-closed instead of warnings?** Because the skill's promise is
 "unbreakable, zero-assumption directive." A warning you can ignore is the
-opposite of idiot-proof. The gate is the product.
+opposite of idiot-proof — so structural violations are fail-closed, no
+exceptions. The heuristic and convention rules (R1, R5, R8, R13) are the
+narrow exception: they report as warnings because they're tripwires, not
+proof, and tripwires can false-positive. Pass `--strict` to make them fail
+the draft too. The gate is still the product.
 
 **Single-service workflow — do I need the routing section?** No. Rule 9 is
 vacuous when no services are named. Delete the Routing Architecture section

{frontface_proofit-1.2.0 → frontface_proofit-1.2.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "frontface-proofit"
-version = "1.2.0"
+version = "1.2.1"
 description = "Fail-closed linter for locked-down runbooks: ten rules of workflow discipline as code"
 readme = "README.md"
 license = { text = "MIT" }
@@ -26,5 +26,8 @@ proofit = "proofit.__main__:main"
 [tool.setuptools.packages.find]
 where = ["src"]
 
+[tool.setuptools.package-data]
+proofit = ["templates/*.md"]
+
 [tool.pytest.ini_options]
 testpaths = ["tests"]

{frontface_proofit-1.2.0 → frontface_proofit-1.2.1/src/frontface_proofit.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: frontface-proofit
-Version: 1.2.0
+Version: 1.2.1
 Summary: Fail-closed linter for locked-down runbooks: ten rules of workflow discipline as code
 Author: Ryan Wesley Hitchcock
 License: MIT
@@ -21,8 +21,10 @@ people following it do the same thing and get the same result. proofit checks
 a runbook draft against fourteen rules of workflow discipline — ten structural
 rules plus the 1337 module's four (cross-role duplication, header date,
 suspicious character count, missing-file verification) — and **rejects the
-draft if any rule is violated**. No warnings, no advisories. Fail-closed:
-exit 1 until it's fixed, exit 0 when it's locked down.
+draft if any structural rule is violated**. Heuristic and convention rules
+(R1, R5, R8, R13) come back as warnings — shown, but they don't fail the
+draft unless `--strict` is passed. Fail-closed where it counts:
+exit 1 until the structural rules pass, exit 0 when it's locked down.
 
 It checks the runbook's **shape, not its truth**: it can tell you a step is
 missing its verification block, not whether the verification would actually
@@ -74,10 +76,11 @@ mechanical: the rules that can be checked as structure *are* checked as
 structure, so compliance isn't a matter of discipline — it's a matter of
 passing the gate.
 
-Two rules (1 and 8) are phrase-list heuristics — tripwires, not proof. Under
-fail-closed they still block, so they are deliberately narrow: they fire only
-on clear hits. A foolproof gate that misfires on good drafts isn't foolproof,
-it's a tax. The narrowness is documented in the code and tested.
+Two rules (1 and 8) are phrase-list heuristics — tripwires, not proof. They
+report as warnings in default mode and block only under `--strict`, so they
+are deliberately narrow: they fire only on clear hits. A foolproof gate that
+misfires on good drafts isn't foolproof, it's a tax. The narrowness is
+documented in the code and tested.
 
 ## Quickstart
 
@@ -85,14 +88,20 @@ Zero dependencies beyond Python 3.10+.
 
 ```bash
 cd proofit
-PYTHONPATH=src python3 -m unittest discover -s tests   # 65 offline tests
+PYTHONPATH=src python3 -m unittest discover -s tests   # 90 offline tests
 ```
 
-Lint a draft:
+Lint a known-good draft first (expect PASS: no violations):
+
+```bash
+PYTHONPATH=src python3 -m proofit lint examples/fixed-runbook.md
+```
+
+The blank template REJECTS until its placeholders are real observable checks. Then lint a draft:
 
 ```bash
 PYTHONPATH=src python3 -m proofit lint my-runbook.md
-# violations listed as RULE n [location] ERROR — message + fix
+# violations listed as RULE n [location] ERROR|WARNING — message + fix
 # exit 1: REJECTED — fix and re-run
 # exit 0: PASS — the runbook is locked down
 ```
@@ -127,9 +136,10 @@ fixed draft passes).
   than producing meaningful per-rule violations. Write it in the template,
   get it checked — anything else isn't a bad runbook to proofit, it's not
   a runbook at all.
-- **Output:** a list of violations (`RULE n [location] ERROR — message`
-  plus a `fix:` line), and an exit code: 1 if any violation exists, 0 if
-  clean. Fail-closed — there is no warning level.
+- **Output:** a list of findings (`RULE n [location] ERROR|WARNING — message`
+  plus a `fix:` line), and an exit code: 1 if any ERROR exists, 0 if
+  clean. Structural violations are fail-closed; heuristic and convention
+  hits are warnings unless `--strict`.
 
 ## Worked example
 
@@ -155,7 +165,11 @@ linter.
 
 **Why fail-closed instead of warnings?** Because the skill's promise is
 "unbreakable, zero-assumption directive." A warning you can ignore is the
-opposite of idiot-proof. The gate is the product.
+opposite of idiot-proof — so structural violations are fail-closed, no
+exceptions. The heuristic and convention rules (R1, R5, R8, R13) are the
+narrow exception: they report as warnings because they're tripwires, not
+proof, and tripwires can false-positive. Pass `--strict` to make them fail
+the draft too. The gate is still the product.
 
 **Single-service workflow — do I need the routing section?** No. Rule 9 is
 vacuous when no services are named. Delete the Routing Architecture section

{frontface_proofit-1.2.0 → frontface_proofit-1.2.1}/src/frontface_proofit.egg-info/SOURCES.txt

@@ -14,6 +14,7 @@ src/proofit/leet.py
 src/proofit/linter.py
 src/proofit/model.py
 src/proofit/parser.py
+src/proofit/templates/runbook-template.md
 tests/test_1337.py
 tests/test_cli.py
 tests/test_ledger.py

{frontface_proofit-1.2.0 → frontface_proofit-1.2.1}/src/proofit/__main__.py

@@ -13,9 +13,25 @@ from . import __version__, lint_file
 from .ledger import Ledger
 
 
-def _template_path() -> str:
+def _template_text() -> str:
+    """Blank runbook template text.
+
+    Packaged copy first (``src/proofit/templates/`` ships in the wheel
+    as package data, so this works from a pip install), repo-layout
+    copy as fallback (``proofit/templates/`` — the file the live demo
+    page fetches from the repo, which must stay at that path).
+    """
+    try:
+        from importlib import resources
+        return (resources.files("proofit") / "templates"
+                / "runbook-template.md").read_text(encoding="utf-8")
+    except (ImportError, OSError):
+        pass
     here = os.path.dirname(os.path.abspath(__file__))
-    return os.path.normpath(os.path.join(here, "..", "..", "templates", "runbook-template.md"))
+    repo_path = os.path.normpath(os.path.join(
+        here, "..", "..", "templates", "runbook-template.md"))
+    with open(repo_path, encoding="utf-8") as f:
+        return f.read()
 
 
 def cmd_lint(path: str, *, strict: bool = False) -> int:
@@ -103,12 +119,10 @@ def cmd_ledger(argv: list[str]) -> int:
 
 
 def cmd_template() -> int:
-    path = _template_path()
     try:
-        with open(path, encoding="utf-8") as f:
-            sys.stdout.write(f.read())
+        sys.stdout.write(_template_text())
     except OSError as e:
-        print(f"error: template not found at {path}: {e}", file=sys.stderr)
+        print(f"error: template not found: {e}", file=sys.stderr)
         return 2
     return 0
 

frontface_proofit-1.2.1/src/proofit/templates/runbook-template.md

@@ -0,0 +1,103 @@
+# Runbook: [workflow name]
+
+<!--
+  Write the draft in this shape, then run:  python -m proofit lint this-file.md
+  Fail-closed: ANY violation rejects the draft until fixed.
+  One action per Do:. Name exact visible labels in **bold**. No new
+  prerequisites after step 1. End with Validation & Closure, never a step.
+-->
+
+## Ground Rules & Prerequisites
+
+### Required Baseline
+
+[Exact starting condition: state, setup, preconditions.]
+
+### Routing Architecture
+
+[Delete this whole section if the workflow uses a single service.]
+
+#### Input spec
+
+[Exact format, structure, and constraints of the input.]
+
+#### Service 1: [name]
+
+- **When to use:** [exact condition or trigger]
+- **Free-tier limits:** [rate limits, context window, monthly quota]
+- **Prompt (verbatim):**
+  ```
+  [paste the exact prompt — no improvisation at runtime]
+  ```
+- **Expected output format:** [schema, field names, data types]
+
+#### Service 2: [name]
+
+- **When to use:** [exact condition or trigger]
+- **Free-tier limits:** [rate limits, context window, monthly quota]
+- **Prompt (verbatim):**
+  ```
+  [paste the exact prompt]
+  ```
+- **Expected output format:** [schema, field names, data types]
+
+#### Rate limit handling
+
+[What to do when a free-tier limit is hit: wait time, fallback service, graceful degradation.]
+
+#### Error paths
+
+[What to do if any service fails or returns unexpected data.]
+
+### Needed Tools/Items
+
+- Hardware: [e.g. a laptop with 8 GB RAM]
+- Software: [e.g. Chrome, Python 3.12]
+- Accounts: [e.g. a free-tier account on each service above — no API keys]
+- Permissions: [e.g. admin on the target machine]
+- Files: [e.g. ~/data/input.csv]
+- Internet access: [e.g. required throughout]
+
+### Expected Outcome
+
+[What will exist after completion, and what it should contain.]
+
+## Step-by-Step Directives
+
+### Step 1: [action name]
+
+**Do:** [exactly ONE action — "Open **Settings**.", never "Open Settings and click Display."]
+
+**Look for:** [exact visual anchor or expected output, including what visibly changes]
+
+**Verify:** [observable success condition — what the screen displays, what exists, what the output contains]
+
+**If wrong:** [the most likely mistake for this step]
+
+**Fix:** [the shortest recovery path — resume the guide instead of restarting when possible]
+
+### Step 2: [action name]
+
+**Do:** [one action]
+
+**Look for:** [visual anchor]
+
+**Verify:** [observable success condition]
+
+**If wrong:** [most likely mistake]
+
+**Fix:** [shortest recovery path]
+
+## Validation & Closure
+
+### Success Condition
+
+[The exact final state. The reader should know with certainty the task succeeded.]
+
+### Consistency Checkpoints
+
+- [ ] All inputs met spec requirements (format, structure, constraints).
+- [ ] Each service received the correct prompt and context.
+- [ ] Each service's output matched the expected format.
+- [ ] Error recovery was never triggered, OR was triggered and resolved correctly.
+- [ ] The complete workflow produces the intended result.

{frontface_proofit-1.2.0 → frontface_proofit-1.2.1}/tests/test_cli.py

@@ -49,6 +49,35 @@ class TestCLI(unittest.TestCase):
         self.assertIn("lint", r.stdout)
 
 
+class TestTemplatePackaging(unittest.TestCase):
+    """The template ships in the wheel (1.2.0 bug: it did not).
+
+    `proofit/templates/runbook-template.md` is the canonical file —
+    the live demo page fetches it from the repo — and
+    `src/proofit/templates/runbook-template.md` is the packaged copy.
+    The two must stay byte-identical, and `proofit template` must
+    print exactly that file, from a pip install as well as from source.
+    """
+
+    def test_packaged_template_matches_repo_template(self):
+        sys.path.insert(0, os.path.join(PKG, "src"))
+        from importlib import resources
+        packaged = (resources.files("proofit") / "templates"
+                    / "runbook-template.md").read_bytes()
+        with open(os.path.join(PKG, "templates",
+                               "runbook-template.md"), "rb") as f:
+            repo = f.read()
+        self.assertEqual(packaged, repo)
+
+    def test_template_command_prints_repo_template(self):
+        r = run_cli("template")
+        self.assertEqual(r.returncode, 0)
+        with open(os.path.join(PKG, "templates",
+                               "runbook-template.md"),
+                  encoding="utf-8") as f:
+            self.assertEqual(r.stdout, f.read())
+
+
 # PARAM-PROOFIT-001: WARNINGs pass by default, fail under --strict;
 # ERRORs fail in both modes.
 R1_ONLY = """# Runbook: test