@amsterdamdatalabs/adl-agent-control-linux-x64 0.1.30 → 0.1.32

package/packages/repo-controls/package/adl-compliance/src/adl_compliance/core/scripts.py

@@ -0,0 +1,120 @@
+"""Which language a code file is in, by name or shebang: one selector for every gate that grades scripts."""
+
+from __future__ import annotations
+
+import re
+from collections.abc import Callable
+
+
+class Names:
+    """The literals the selector reads, on a class like every gate's."""
+
+    TYPESCRIPT_SUFFIXES = (".ts", ".mts", ".cts")
+    SLASH_SUFFIXES = (*TYPESCRIPT_SUFFIXES, ".tsx", ".js", ".jsx", ".mjs", ".cjs")
+    RUST_SUFFIX = ".rs"
+    PYTHON_SUFFIX = ".py"
+    SHELL_SUFFIXES = (".sh", ".bash")
+    ZSH_SUFFIX = ".zsh"
+    JUSTFILE_SUFFIXES = (".just", ".justfile")
+    JUSTFILES = ("justfile", ".justfile")
+
+    SHEBANG_POSIX = frozenset({"sh", "dash"})
+    SHEBANG_BASH = "bash"
+    SHEBANG_KSH = frozenset({"ksh", "ksh93", "mksh"})
+    SHEBANG_SPLIT = "--split-string="
+    SHEBANG_ZSH = "zsh"
+    SHEBANG_NODES = frozenset({"node", "bun", "deno", "tsx", "ts-node"})
+    SHEBANG_PYTHON = re.compile(r"python[0-9.]*|uv")
+    SHEBANG_ENV_VALUED = frozenset({"-u", "--unset", "-C", "--chdir"})
+
+    SHELL_LANGUAGES = frozenset({"shell", "posix", "mksh", "zsh", "justfile"})
+    DIALECT_SUFFIXES = (".sh", ".bash")
+
+
+def language(relative: str, text: str) -> str | None:
+    """`python`, `rust`, `typescript`, `slash`, `shell` (bash), `posix`, `mksh`, `zsh`, `justfile`, or None when the file is none of them."""
+    name = relative.rsplit("/", 1)[-1]
+    if relative.endswith(Names.PYTHON_SUFFIX):
+        return "python"
+    if relative.endswith(Names.RUST_SUFFIX):
+        return "rust"
+    if relative.endswith(Names.TYPESCRIPT_SUFFIXES):
+        return "typescript"
+    if relative.endswith(Names.SLASH_SUFFIXES):
+        return "slash"
+    if name.lower() in Names.JUSTFILES or relative.endswith(Names.JUSTFILE_SUFFIXES):
+        return "justfile"
+    if relative.endswith(Names.SHELL_SUFFIXES):
+        declared = _shebang_language(text)
+        return declared if declared in Names.SHELL_LANGUAGES else "shell"
+    if relative.endswith(Names.ZSH_SUFFIX):
+        return "zsh"
+    if "." in name:
+        return None
+    return _shebang_language(text)
+
+
+def _shebang_language(text: str) -> str | None:
+    first = text.split("\n", 1)[0].rstrip("\r")
+    if not first.startswith("#!"):
+        return None
+    words = first[2:].split()
+    if not words:
+        return None
+    program = words[0].rsplit("/", 1)[-1]
+    if program == "env":
+        rest = _env_command(words[1:])
+        if not rest:
+            return None
+        program = rest[0].rsplit("/", 1)[-1]
+    if program in Names.SHEBANG_POSIX:
+        return "posix"
+    if program == Names.SHEBANG_BASH:
+        return "shell"
+    if program in Names.SHEBANG_KSH:
+        return "mksh"
+    if program == Names.SHEBANG_ZSH:
+        return "zsh"
+    if program in Names.SHEBANG_NODES:
+        return "slash"
+    if Names.SHEBANG_PYTHON.fullmatch(program):
+        return "python"
+    return None
+
+
+def _env_command(words: list[str]) -> list[str]:
+    """The command `env` runs: its own options, their values and NAME=value words skipped."""
+    rest: list[str] = []
+    skip = False
+    for word in words:
+        if skip:
+            skip = False
+        elif rest:
+            rest.append(word)
+        elif word in Names.SHEBANG_ENV_VALUED:
+            skip = True
+        elif word.startswith(Names.SHEBANG_SPLIT):
+            rest.append(word[len(Names.SHEBANG_SPLIT) :])
+        elif word.startswith("-S") and len(word) > 2:
+            rest.append(word[2:])
+        elif not word.startswith("-") and "=" not in word:
+            rest.append(word)
+    return rest
+
+
+def _selects(relative: str, head: Callable[[], str | None] | None, wanted: frozenset[str] | None) -> bool:
+    name = relative.rsplit("/", 1)[-1]
+    found = language(relative, "")
+    if found is None and "." not in name and head is not None:
+        found = _shebang_language(head() or "")
+    return found is not None and (wanted is None or found in wanted)
+
+
+def is_code_file(relative: str, head: Callable[[], str | None] | None = None) -> bool:
+    """Whether the file is code in a language this estate grades; an extensionless file is read for its shebang."""
+    return _selects(relative, head, None)
+
+
+def is_shell_script(relative: str, head: Callable[[], str | None] | None = None) -> bool:
+    """Whether the file is a shell, zsh or just script, by name or shebang."""
+    return _selects(relative, head, Names.SHELL_LANGUAGES)

package/packages/repo-controls/package/adl-compliance/src/adl_compliance/policy/config.py

@@ -1,6 +1,6 @@
 """`<name>` in a scope pattern, and the one file that says what it means.
 
-A row declares what its rule is ABOUT — `<rust>/crates/**/src/*.rs` — and
+A row declares what its rule is ABOUT — `<docs>/**/*.md` — and
 `trees.toml` says where that currently lives. The registry therefore contains no
 consumer path at all, which is K-42: a path outside this package, written inside
 it, is a dependency on this particular checkout's shape.

package/packages/repo-controls/package/adl-compliance/src/adl_compliance/policy/config.toml

@@ -1,9 +1,9 @@
 # The consumer's layout, declared once.
 #
 # A registry row names a LOGICAL tree and never a physical path, so
-# `<rust>/crates/**/src/*.rs` is a rule about Rust crates and this file is the
-# only place that says Rust lives in `backend/`. Move the tree, edit one line,
-# and all 82 backend-rooted scope entries follow.
+# `<docs>/**/*.md` is a rule about documentation and this file is the
+# only place that says documentation lives in `docs/`. Move the tree, edit one line,
+# and every docs-rooted scope entry follows.
 #
 # This closes K-42 for the registry. A path outside this package written inside
 # it is a dependency on THIS checkout's shape; the rule is that the consumer's
@@ -12,7 +12,7 @@
 # line here, not 79 edits across the rows.
 #
 # SYNTAX IS `<name>`, NOT `{name}`. Braces are already glob alternation and 11
-# rows use it -- `{docs,.plans,.specs}/**`, `{schema,compliance}/**`. A `{rust}`
+# rows use it -- `{docs,plans}/**`, `{schema,compliance}/**`. A `{rust}`
 # placeholder would be indistinguishable from a one-element alternation, and
 # resolving that by "a placeholder has no comma" is the silent-ambiguity trap
 # K-13 records. Angle brackets appear in no scope entry and in no path.

package/packages/repo-controls/package/adl-compliance/src/adl_compliance/policy/gate-registry.alt.yaml

@@ -74,19 +74,19 @@ rows:
         # Same reasoning, the TypeScript-shaped colocated test file: this scope
         # excluded `test_*.py` but nothing that names a TypeScript test the way
         # every managed repo's own convention does. Caught live against
-        # enact-vps-m5: `global-tools/test/install-embedding-model.test.ts`
+        # a managed repository: `global-tools/test/install-embedding-model.test.ts`
         # asserts that `selectVocabulary` DROPS non-Latin-script tokens, and
         # embeds a CJK literal as the negative-test input -- exactly
         # the "test asserting a non-English string" case above, just in `.ts`
         # rather than `.py`. COVERAGE COST, stated not buried: 548 colocated
         # `*.test.ts`/`*.test.tsx` files across the four managed repos
-        # (enact-m5 400, enact-vps-m5 134, enact-repo-controls-m5 14,
-        # enact-extensions 0) stop being graded for english-only.
+        # (400, 134, 14 and 0 in the four
+        # repositories) stop being graded for english-only.
         - '!**/*.test.ts'
         - '!**/*.test.tsx'
         # The four exclusions removed here named
-        # `<typescript>/packages/desktop/...` and `<typescript>/tests/unit/...`
-        # -- one repository's files written into the shared package. Measured
+        # one repository's application and unit-test files, written into the
+        # shared package under a tree placeholder. Measured
         # before removing: 0 of the 4 exist in ANY of the four managed
         # repositories, so they excluded nothing anywhere. If a consumer ever
         # has legitimately non-English source, it declares that itself.
@@ -363,7 +363,6 @@ rows:
       scope:
         - '{AGENTS,CLAUDE}.md'
         - '**/{AGENTS,CLAUDE}.md'
-        - <compliance>/checks/layout/agents-layout/check.sh
         - '!**/fixtures/**'
     grants: 0
     timeout_ms: 60000
@@ -386,7 +385,7 @@ rows:
         #
         # Same shape as LAY-014's `!<workflows>/scripts/**/fixtures/**`, one
         # level up. COVERAGE COST, stated rather than buried: this drops 19 of
-        # enact-m5's 44 `workflows/scripts/` files (43%) from the gate. The risk
+        # one managed repository's 44 `workflows/scripts/` files (43%) from the gate. The risk
         # it trades away is small — a TEST that truly needs a gitignored file
         # fails loudly in CI, whereas a production CI script referencing one can
         # silently skip, and production scripts are all still graded.
@@ -410,29 +409,14 @@ rows:
       # recorded state the register file is deleted. No ceilings declared
       # means nothing to enforce — the gate is inert.
       scope:
-        - <typescript>/**
-        - <rust>/**
-        - <infra>/**
-        - <dev>/**
-        - <workflows>/**
-        # A single '*' cannot cross a '/': it is one path segment, not "the rest
-        # of the subtree". Confirmed live against this checkout, not just the
-        # ported test fixture — locales/de/common.json sits two segments below
-        # locales/, and real crate/build/terraform files sit at varying depths
-        # below their prefix (`rg`/`find` measured 2 and 4 segments for
-        # Cargo.toml, 4 for build.rs, 2 through 5 for main.tf) — every nested
-        # occurrence was silently unexcluded until these five became '**'.
-        - '!<typescript>/packages/desktop/src/renderer/services/i18n/locales/**'
-        - '!<rust>/crates/enact-app/assets/builtin-skills/**'
-        - '!<rust>/crates/enact-app/assets/builtin-harnesses/*'
-        - '!<dev>/seed/access/*'
-        - '!<dev>/seed/infra/*'
-        - '!<dev>/seed/tenancy/*'
-        - '!<dev>/scripts/*'
-        - '!<dev>/ops/*'
-        - '!**/Cargo.toml'
-        - '!**/build.rs'
-        - '!**/main.tf'
+        # The WHOLE repository (Tarun, 2026-10-08, PBI 598): the ratchet is
+        # repo-wide and hardcodes no exclusion. `**` matches every path of the
+        # Git-visible universe the CLI hands over (`ls-files --cached --others
+        # --exclude-standard`), and the measurement itself skips dot-prefixed
+        # components. What a repository exempts it declares itself: a ceiling
+        # per tree in [controls.compliance.directory-ratchet].max_children and
+        # its over-limit / single-file baselines in the ratchet register below.
+        - '**'
         # The ratchet register is a named INPUT of this gate, not a scope entry:
         # it is the CONSUMER's own file (<compliance>/ratchets/
         # LAY-004-over-limit-directory.yaml -- the estate's settled location for
@@ -485,21 +469,21 @@ rows:
   - id: LAY-008
     name: placement
     domain: layout
-    title: Run retained placement gate
+    title: Reject throwaway files and undeclared image placement
     severity: M
     controls:
       - EC-A.5.36
     engine:
+      # Every file: a throwaway extension is reported anywhere, and an image or
+      # screenshot-named file is reported unless it sits in a directory the
+      # repository declares in `asset-dirs`. (The root allowlist is LAY-012's
+      # `root-closed-set`, not this row's.)
       scope:
-        - '*'
-        - '*.{log,tmp,bak,swp}'
-        - '**/*.{log,tmp,bak,swp}'
-        - <compliance>/checks/layout/placement/check.sh
+        - '**'
         # The corpora under placement-allowed/ and placement-stray/ are the
-        # gate's own test input: enumerated asset paths, a screenshot, a
-        # markdown file inside a crate src/ tree, and a file under retired
-        # apps/. Graded as production content they are exactly the findings
-        # the gate exists to raise.
+        # gate's own test input: a screenshot and a throwaway file. Graded as
+        # production content they are exactly the findings the gate exists to
+        # raise.
         - '!**/fixtures/**'
     grants: 0
     timeout_ms: 60000
@@ -567,7 +551,6 @@ rows:
         - '!.git/**'
         - '!.DS_Store'
         - '!.cache/**'
-        - '!.enactrs/**'
         - '!.pytest_cache/**'
         - '!.ruff_cache/**'
         - '!.scratch/**'
@@ -652,27 +635,6 @@ rows:
     grants: 0
     timeout_ms: 60000
 
-  # ---- pipeline ----
-  - id: PIP-004
-    name: phase-disjoint
-    domain: pipeline
-    title: Enforce phase mutual exclusion across every registry row
-    severity: M
-    controls:
-      - EC-A.5.36
-    engine:
-      # Scope is SELECTION, not resolution. The check finds the registry
-      # structurally, from its own position in the package, and that is why it
-      # needs no path here to READ it -- but scope decides whether a changed
-      # file makes this row run at all, and an empty scope matches nothing. The
-      # gate that enforces phase disjointness across the registry was therefore
-      # never selected by a change to the registry: the one edit that can break
-      # it was the one edit that did not run it.
-      scope:
-        - <compliance>/policy/gate-registry.yaml
-    grants: 0
-    timeout_ms: 60000
-
   # ---- repository ----
   - id: REP-001
     name: license-declared
@@ -707,53 +669,32 @@ rows:
       - EC-A.5.21
     engine:
       scope:
-        # The tree-derived DEFAULT: each half is exactly two entries under its own
-        # tree. A repository whose manifests do not sit one level under its tree
-        # (enact-combined: a brace-union `typescript` tree, manifests at the root and
-        # under apps/*/) states the paths itself with `[controls.compliance.lockfiles]`
-        # (`bun-manifests`/`bun-lockfiles`, `cargo-manifests`/`cargo-lockfiles`); a
-        # declared half REPLACES that half's two entries, in the scope and in what the
-        # check grades. See `LockfileIntegrity.declared_scope`.
-        - <rust>/Cargo.toml
-        - <rust>/Cargo.lock
-        - <typescript>/package.json
-        - <typescript>/bun.lock
+        # The REPOSITORY-ROOT pairs, graded automatically (Tarun, 2026-10-08,
+        # PBI 598): `Cargo.toml` -> `Cargo.lock` and `package.json` -> `bun.lock`.
+        # Every other lockfile is graded only because the repository declares it
+        # with `[controls.compliance.lockfiles]` (`bun-manifests`/`bun-lockfiles`,
+        # `cargo-manifests`/`cargo-lockfiles`); a declaration ADDS its manifests
+        # and lockfiles to this scope and to what the check grades. See
+        # `LockfileIntegrity.declared_scope`.
+        - Cargo.toml
+        - Cargo.lock
+        - package.json
+        - bun.lock
     grants: 0
     timeout_ms: 60000
   - id: REP-004
-    name: no-banned-commands
+    name: no-legacy-search
     domain: repository
-    title: Forbid banned command invocations in shell scripts and justfiles
+    title: Forbid standalone legacy search in shell scripts and justfiles
     severity: M
     controls:
       - EC-A.5.36
     engine:
       scope:
-        # Both GIT-001 (no version-control invocation) and GRP-001 (no
-        # standalone grep, use rg) are restricted to the compliance package --
-        # a deliberate, explicit decision (both are compliance gates), not an
-        # inherited default. Neither rule is about the rest of the repository,
-        # so nothing here is expressed as an extension glob any more: the old
-        # repository-wide `*.{sh,bash,...}` list existed only because GRP-001
-        # used to run everywhere and needed to exclude non-shell files from a
-        # much larger candidate set. Restricted to compliance/**, that ceiling
-        # is gone -- check.sh's own shell/justfile filter is enough.
-        - <compliance>/**
-        # The scope contract itself. `discover_files` lists the tree through the
-        # version-control plumbing; every other file in this package reaches it
-        # through this one, which is why it is exempted here rather than at each
-        # caller. Out of scope, not granted: the rule does not apply to the
-        # implementation of the mechanism it exists to make optional everywhere
-        # else.
-        - '!<compliance>/lib/discovery.sh'
-        # The same exemption, on the test side. `_corpus-set.sh` is the one
-        # place a fixture enumerates its corpus, and it asks the corpus the
-        # same question the runner asks the repository, through the same
-        # plumbing -- deliberately, so the set a test hands over is the same
-        # KIND of set production hands over. Every converted gate test reaches
-        # it through this file, which is why it is exempted here rather than at
-        # each caller.
-        - '!<compliance>/checks/framework/_corpus-set.sh'
+        # Repo-wide: every file, narrowed by the check to the shell scripts
+        # and justfiles `core/scripts.py` selects (by suffix, name or
+        # shebang), the same selector REP-007 uses.
+        - '**'
         # EVERY gate's fixture corpus, not just this row's. A fixture is
         # deliberate violation material -- both halves of this row's own corpus
         # contain real invocations on purpose -- and graded as production
@@ -775,3 +716,56 @@ rows:
         - '**'
     grants: 0
     timeout_ms: 60000
+  # Split from REP-004 (Tarun 2026-10-08: PBI 605). Same repo-wide scope; a
+  # repository whose git hooks and release scripts must call the version-control
+  # tool leaves this gate out of `enabled` and keeps REP-004.
+  - id: REP-006
+    name: no-vcs-in-scripts
+    domain: repository
+    title: Forbid version-control calls in shell scripts and justfiles
+    severity: M
+    controls:
+      - EC-A.5.36
+    engine:
+      scope:
+        # Repo-wide: every file, narrowed by the check to the shell scripts
+        # and justfiles `core/scripts.py` selects (by suffix, name or
+        # shebang), the same selector REP-007 uses.
+        - '**'
+        # EVERY gate's fixture corpus, not just this row's. A fixture is
+        # deliberate violation material -- both halves of this row's own corpus
+        # contain real invocations on purpose -- and graded as production
+        # content it fails the gate it exists to prove. Declared here rather
+        # than skipped inside the check: path exclusions carry no default in
+        # this tree.
+        - '!**/fixtures/**'
+        # The managed projections adl rebuilds on every install (Tarun,
+        # 2026-10-08): their git hooks must call the version-control tool and
+        # the repository cannot edit them. Only this gate leaves them out.
+        - '!.workflows/**'
+        - '!.agents/**'
+        - '!.claude/**'
+        - '!.codex/**'
+        - '!.cursor/**'
+        - '!.kimi-code/**'
+        - '!.opencode/**'
+    grants: 0
+    timeout_ms: 60000
+  # Bug 602 (Tarun 2026-10-08): no multi-line comment in code; the explanation
+  # moves to docs/. Grades every code file by suffix (the check's own question,
+  # `scanner.is_code_file`), so the scope is the whole repository less every
+  # gate's fixture corpus. Exempt files are named by the repository's own
+  # register, `<compliance>/ratchets/REP-007-multiline-comments.yaml`.
+  - id: REP-007
+    name: no-multiline-comments
+    domain: repository
+    title: Forbid multi-line comments in code files
+    severity: M
+    controls:
+      - EC-A.8.28
+    engine:
+      scope:
+        - '**'
+        - '!**/fixtures/**'
+    grants: 0
+    timeout_ms: 60000