claude-dev-env 2.8.0 → 2.10.0

package/agents/code-quality-agent.md

@@ -80,7 +80,7 @@ For reusable Variant C audit prompts scoped to a single category, see `../audit-
 }
 ```
 
-`id` uses the form `loop<N>-<K>` for /bugteam and /qbug invocations and `find<K>` for /findbugs. The orchestrator supplies the prefix in the prompt; honor whatever it gives you.
+`id` uses the form `loop<N>-<K>` for /bugteam and pr-converge invocations and `find<K>` for standalone audit calls. The orchestrator supplies the prefix in the prompt; honor whatever it gives you.
 
 **The `failure_mode` field is the audit-to-fix handoff.** State the failing line, the desired post-fix property, and a one-line validation the fix agent can run to confirm correctness. The fix agent reads `failure_mode` without re-running your audit — make it self-sufficient.
 
@@ -113,6 +113,14 @@ A bare verified-clean label is inadequate: every Shape B entry lists the files o
 | P1 | Regression, silent failure, or behavior change that escapes existing tests. |
 | P2 | Dead code, minor smell, style issue, category J finding without runtime impact. |
 
+## Collection before filtering
+
+Report every real finding at its true severity. Collection retains P0, P1, and
+P2 findings with file, line, evidence (`excerpt` / `failure_mode`), and
+category. Do not drop lower-severity real findings during collection so a later
+consumer can filter. Severity or action filtering is a separate stage after the
+collection record is complete.
+
 ## Per-Category Expectation
 
 Every category A through Q is investigated. The output for each category is one of:
@@ -175,7 +183,7 @@ Followed by the Shape A finding list, the Shape B proof-of-absence list, and the
 
 ## Caller Context
 
-Callers /bugteam, /qbug, and /findbugs invoke this agent at different models per call (opus for /bugteam, sonnet primary for /findbugs, haiku secondary for both /qbug and /findbugs). The frontmatter carries no `model:` key, so each caller's `Agent()` model applies. Persistence files such as `loop-N-audit.json` and `loop-N-diagnostics.json` are the calling skill's responsibility — your output is the structured finding list defined above.
+Callers /bugteam, /pr-converge, and /autoconverge invoke this agent at different models per call (opus for /bugteam; the PR-loop orchestrators set their own Agent model). The frontmatter carries no `model:` key, so each caller's `Agent()` model applies. Persistence files such as `loop-N-audit.json` and `loop-N-diagnostics.json` are the calling skill's responsibility — your output is the structured finding list defined above.
 
 ## Examples
 

package/agents/code-verifier.md

@@ -1,6 +1,6 @@
 ---
 name: code-verifier
-description: Post-hoc verification agent for the three-phase code workflow. Spawned by the main session after coder agents finish. Runs every check itself in a fresh context — named gates, tests against recorded baselines, two-way diff-vs-task reading — puts the draft verdict through one strongest-tier validation subagent that tries to refute it, then ends with a fenced verdict block the verifier_verdict_minter hook turns into the commit-gate verdict. Never edits files in the tree under review — its one exception is a deliberate break at an off-tree break site outside that tree, defined in its body.
+description: Post-hoc verification agent. Runs only when the task owner asks for a verification pass — never spawned automatically because a coder agent finished. Runs every check itself in a fresh context — named gates, tests against recorded baselines, two-way diff-vs-task reading — puts the draft verdict through one strongest-tier validation subagent that tries to refute it, then ends with a fenced verdict block the verifier_verdict_minter hook turns into the commit-gate verdict. Never edits files in the tree under review — its one exception is a deliberate break at an off-tree break site outside that tree, defined in its body.
 tools: Read, Grep, Glob, Bash, Task
 color: orange
 ---

package/agents/test_agent_frontmatter.py

@@ -23,23 +23,20 @@ a plain scalar reads as a mapping key and makes the whole block unloadable::
     flag: description: ... constraints. Examples:   <- block no longer loads
 
 Two parsers read these files, and a block that satisfies one can still be
-unreadable by the other. `scripts/codex_compat_materializer.py` reads the
-block line by line, so every field fits on the line that names it::
+unreadable by the other. `scripts/codex_compat_materializer.py` loads the
+block through YAML with duplicate-key rejection, so block-scalar descriptions
+and empty `tools: []` lists parse::
 
-    ok:   description: "Use this agent ... Examples:\\n\\n  <example> ..."
-    flag: description: |        <- the line scan cannot follow the block scalar
+    ok:   description: |
+            multi-line body
+    ok:   tools: []
 
-That reader also counts fence lines across the whole file, so a bare `---`
+That reader still counts fence lines across the whole file, so a bare `---`
 anywhere in the body reads as a second frontmatter fence::
 
     ok:   ```yaml ... ``` example blocks in the body
     flag: a `---` line inside a body example  <- counted as a fence
 
-That reader also cannot express an empty `tools: []` list, which is a defect
-in the reader rather than in the one definition that declares it, so that
-definition is named in `_codex_materializable_paths` and covered by every
-other check here.
-
 Each definition also has to carry a `name` and a `description` bound to a
 non-empty string, and a `name` equal to its file stem — a mapping that loads
 but binds `description` to nothing, or names an agent the file does not,
@@ -82,7 +79,6 @@ import pytest
 import yaml
 
 ACCEPTED_FRONTMATTER_KEYS = frozenset({"name", "description", "tools", "color"})
-EMPTY_TOOLS_LIST_FILENAME = "code-advisor.md"
 EXEMPT_MARKDOWN_FILENAME = "CLAUDE.md"
 FRONTMATTER_FENCE_LINE = "---"
 MATERIALIZER_MODULE_NAME = "codex_compat_materializer"
@@ -156,36 +152,9 @@ def _top_level_keys(frontmatter_block: str) -> set[str]:
     return set(TOP_LEVEL_KEY_PATTERN.findall(frontmatter_block))
 
 
-@cache
-def _codex_materializable_paths() -> tuple[Path, ...]:
-    """Return the definitions the Codex materializer is expected to read.
-
-    `code-advisor.md` declares `tools: []`, and its body states the agent has
-    zero tools, so the empty list is the field saying what the author meant.
-    The materializer's list parser raises on an empty list, so it cannot
-    express a correct declaration::
-
-        tools: Read, Bash  -> ok:   parsed
-        tools: []          -> flag: MaterializerError, though the file is right
-
-    The gap belongs to that parser, so the definition stays as written and only
-    this one check steps around it. The YAML-load, required-field, accepted-key,
-    model-ban, and name checks all still cover the file.
-
-    Returns:
-        Every agent definition path except the one whose correct frontmatter
-        the materializer's list parser cannot express.
-    """
-    return tuple(
-        each_path
-        for each_path in _agent_definition_paths()
-        if each_path.name != EMPTY_TOOLS_LIST_FILENAME
-    )
-
-
 @cache
 def _codex_materializer_module() -> ModuleType:
-    """Load the package's own line-oriented frontmatter parser from disk.
+    """Load the package's YAML frontmatter materializer from disk.
 
     Returns:
         The imported `codex_compat_materializer` module.
@@ -317,6 +286,29 @@ def test_agent_frontmatter_loads_as_a_yaml_mapping(
     )
 
 
+@pytest.mark.parametrize(
+    "agent_file_name",
+    (
+        "docs-agent.md",
+        "issue-tracker.md",
+        "skill-writer-agent.md",
+    ),
+)
+def named_agents_yaml_safe_load_as_mapping(agent_file_name: str) -> None:
+    """P-107 regression: named agents remain real YAML mappings under safe_load."""
+    agent_definition_path = Path(__file__).parent / agent_file_name
+    assert agent_definition_path.is_file(), (
+        f"{agent_file_name} missing from agents/ — P-107 surface gone"
+    )
+    parsed_frontmatter = yaml.safe_load(_frontmatter_block(agent_definition_path))
+    field_problem = _required_field_problem(parsed_frontmatter)
+    assert field_problem is None, f"{agent_file_name} {field_problem}"
+    name_problem = _agent_name_problem(
+        parsed_frontmatter, agent_definition_path.stem
+    )
+    assert name_problem is None, f"{agent_file_name} {name_problem}"
+
+
 def test_every_agent_definition_yields_a_frontmatter_block() -> None:
     covered_names = {each_path.name for each_path in _agent_definition_paths()}
     uncovered_names = sorted(
@@ -333,7 +325,7 @@ def test_every_agent_definition_yields_a_frontmatter_block() -> None:
 
 @pytest.mark.parametrize(
     "agent_definition_path",
-    _codex_materializable_paths(),
+    _agent_definition_paths(),
     ids=lambda each_path: each_path.name,
 )
 def test_agent_frontmatter_parses_with_the_codex_materializer(
@@ -422,3 +414,29 @@ def test_agent_frontmatter_carries_no_model_key(
         "the caller supplies the model on every spawn, so agent definitions "
         "carry no model key at all, not even model: inherit"
     )
+
+
+def _clean_coder_body() -> str:
+    return (Path(__file__).parent / "clean-coder.md").read_text(encoding="utf-8")
+
+
+def test_clean_coder_never_globs_or_reads_dotenv_files() -> None:
+    body = _clean_coder_body()
+    assert "`**/.env`" not in body
+    assert "`**/.env.*`" not in body
+    assert "Never open `.env`" in body or "Do **not** glob or open `.env`" in body
+    assert re.search(r"(?i)glob.*\.env|\.env.*glob", body) is None or "Do **not** glob or open `.env`" in body
+
+
+def test_clean_coder_uses_task_local_config_discovery() -> None:
+    body = _clean_coder_body()
+    assert "task-local" in body.lower()
+    assert "project-wide preload" in body.lower() or "Do **not** glob the whole tree" in body
+    assert "Issue all five Glob calls" not in body
+    assert "Issue all seven Glob calls" not in body
+
+
+def test_clean_coder_examples_import_constants_from_config() -> None:
+    body = _clean_coder_body()
+    assert "from config.timing import MAXIMUM_RETRIES" in body
+    assert re.search(r"(?m)^MAXIMUM_RETRIES\s*=\s*\d+", body) is None

package/audit-rubrics/CLAUDE.md

@@ -6,6 +6,7 @@ Audit rubrics for the PR-loop code-review suite. The rubrics define the 17 bug c
 
 | File | Purpose |
 |---|---|
+| `audit-categories.json` | Machine-readable A-Q schema: id, title, slug, and sub-bucket id/axis pairs; single source for rubric and prompt skeleton parity |
 | `source-material-section-types.md` | Lookup table for how to chunk an artifact into sections for an audit prompt; covers code PRs, docs, SQL schemas, config files, and more |
 
 ## Subdirectories
@@ -39,4 +40,4 @@ Audit rubrics for the PR-loop code-review suite. The rubrics define the 17 bug c
 
 ## Breaking-change rule
 
-Adding a sub-bucket to a category rubric requires updating the matching prompt template in `prompts/` in the same commit. Skills that reference category IDs (`bugteam`, `findbugs`) rely on stable sub-bucket IDs (A1, A2, … P-n).
+Adding a sub-bucket to a category rubric requires updating `audit-categories.json` and the matching prompt template in `prompts/` in the same commit, then running `audit_category_schema.py --validate`. Skills that reference category IDs (`bugteam`, `findbugs`) rely on stable sub-bucket IDs (A1, A2, … Q-n). Worked examples stay in the rubric markdown only; they are outside the schema.