agentforge-framework 0.2.4__tar.gz → 0.2.5__tar.gz

{agentforge_framework-0.2.4 → agentforge_framework-0.2.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agentforge-framework
-Version: 0.2.4
+Version: 0.2.5
 Summary: A modular framework for coordinating specialized software agents.
 Author: Yash Mhatre
 License-Expression: Apache-2.0
@@ -26,9 +26,9 @@ A human states a Task. The Orchestrator grills them on it, writes a spec, cuts t
 
 ## Status
 
-The Workflow runtime now runs multiple Roles in order. The default `feature`
-Workflow invokes the Implementer, the Tester, Security, and then the Reviewer,
-posting each Agent Result to the Issue before starting the next Step.
+A Workflow runs its Roles in order. The default `feature` Workflow invokes the
+Implementer, the Tester, Security, and then the Reviewer, posting each Agent
+Result to the Issue before starting the next Step.
 
 ```console
 $ agentforge plan "add a retry to the loader"
@@ -104,13 +104,15 @@ cannot hold it — `codex exec` discards its approval flag — so a denied Run o
 `codex` is refused before it starts rather than run under a guarantee nothing
 enforces (ADR-0007, amended). Security, the Reviewer, and the Architect need no such flag —
 auditing, reviewing, and designing are reading. All six Roles `CONTEXT.md` names
-now run. Plugins have landed, and with them the data-engineering ones: a
+run. Plugins carry the data-engineering conventions: a
 repository whose files import `pyspark` has its Roles told to write DataFrame
 expressions rather than RDDs, and a Databricks workspace gets Unity Catalog
 naming and the Delta MERGE idioms in the prompts that write code and the
 workspace's own posture in the one that audits it. A repository that matches
-neither hears neither. `agentforge init` is still to come. See
-[`docs/PLAN.md`](docs/PLAN.md).
+neither hears neither. `agentforge init` inspects a repository and writes its
+`.agentforge/config.yaml`; the file is optional, and
+[Project configuration](#project-configuration) below says what it holds. What
+is still to come is in [`docs/PLAN.md`](docs/PLAN.md).
 
 Before the first Role is invoked, AgentForge resolves a Context Pack from the
 frozen plan — the files it names, the symbols and imports inside them, the
@@ -148,7 +150,7 @@ without notice. See
 ```console
 $ pip install agentforge-framework
 $ agentforge --version
-agentforge 0.2.4
+agentforge 0.2.5
 ```
 
 That is the whole of it. The two routes below the naming note are for a machine
@@ -172,7 +174,7 @@ from the [latest release](https://github.com/yashmhatre/AgentForge/releases/late
 and install the file:
 
 ```console
-$ pip install agentforge_framework-0.2.4-py3-none-any.whl
+$ pip install agentforge_framework-0.2.5-py3-none-any.whl
 ```
 
 Publishing a release is a wizard rather than a checklist:
@@ -209,7 +211,7 @@ AgentForge is a terminal tool. It runs fine from an IDE's terminal, but the IDE
 
 **Antigravity IDE cannot be a Provider** ([#101](https://github.com/yashmhatre/AgentForge/issues/101)). Checked by running it, so nobody has to derive it again: `antigravity-ide chat` accepts a prompt and reads stdin, which is the right shape for the argument vector — but it is a window launcher. Piping a prompt to it returns exit 0 immediately with `Reading from stdin via: …\code-stdin-XXXX` on stdout and nothing else, ever; the answer goes to a GUI session. A Provider adapter has to read the Agent's result off stdout, so there is nothing for `parse_output` to parse. The same is true of any editor CLI of this shape.
 
-**Do not run another agent against a checkout while a Run is going.** A Run commits every change to a file git already tracks, however it got there ([ADR-0015](docs/adr/0015-a-run-commits-what-it-declared.md)) — so a second agent's half-finished edits are committed into the Run's branch and attributed to a Role. Files nothing in the Run claimed are listed in the pull request body ([ADR-0023](docs/adr/0023-a-commit-names-what-nothing-claimed.md)), which makes it visible at Sign-off but does not prevent it. Use a separate clone.
+**Do not run another agent against a checkout while a Run is going.** A Run commits every change to a file git already tracks, however it got there ([ADR-0015](docs/adr/0015-a-run-commits-what-it-declared.md)) — so a second agent's half-finished edits are committed into the Run's branch and attributed to a Role. Files nothing in the Run claimed are listed in the pull request body ([ADR-0023](docs/adr/0023-a-commit-names-what-nothing-claimed.md)), which makes it visible at Sign-off but does not prevent it. Use a separate clone. A second `agentforge implement` is refused outright rather than merely disclosed — a Run holds its checkout for the duration ([ADR-0026](docs/adr/0026-a-run-holds-the-checkout-it-was-pointed-at.md)) — but nothing else consults that lock, so an IDE assistant is still yours to keep away.
 
 ## Commands
 
@@ -277,9 +279,21 @@ if the repository has no GitHub remote, because ADR-0002 makes that a
 precondition for every Run. Re-running never clobbers a config you have edited:
 it reports what differs and writes nothing, and `--force` replaces it.
 
+A Python suite is written pinned to the interpreter that will run it —
+`[".venv/bin/python", "-m", "pytest"]` rather than `pytest` — whenever the
+repository has a virtualenv to pin it to, and init checks that the interpreter
+can run pytest before writing the line. Bare `pytest` is whichever one PATH
+answers with, which in a project with a venv is the one the project does not
+use: the suite then fails on imports that are installed and reads as a broken
+repository rather than a misconfigured one. The interpreter is named relative to
+the repository root, so the file is still right on somebody else's clone, and
+init looks up to two directories down for the project — a suite that lives in
+`subproject/tests/` is found, and named.
+
 The file is not a precondition. Without one, the documented Provider capability
 defaults are Claude `native` and every other Provider `fragment`, and the
-`tests` Gate runs `pytest`.
+`tests` Gate runs `pytest` — the case the pinning above exists to avoid, so a
+repository with a venv is better off with a file.
 
 ```yaml
 providers:

{agentforge_framework-0.2.4 → agentforge_framework-0.2.5}/README.md

@@ -8,9 +8,9 @@ A human states a Task. The Orchestrator grills them on it, writes a spec, cuts t
 
 ## Status
 
-The Workflow runtime now runs multiple Roles in order. The default `feature`
-Workflow invokes the Implementer, the Tester, Security, and then the Reviewer,
-posting each Agent Result to the Issue before starting the next Step.
+A Workflow runs its Roles in order. The default `feature` Workflow invokes the
+Implementer, the Tester, Security, and then the Reviewer, posting each Agent
+Result to the Issue before starting the next Step.
 
 ```console
 $ agentforge plan "add a retry to the loader"
@@ -86,13 +86,15 @@ cannot hold it — `codex exec` discards its approval flag — so a denied Run o
 `codex` is refused before it starts rather than run under a guarantee nothing
 enforces (ADR-0007, amended). Security, the Reviewer, and the Architect need no such flag —
 auditing, reviewing, and designing are reading. All six Roles `CONTEXT.md` names
-now run. Plugins have landed, and with them the data-engineering ones: a
+run. Plugins carry the data-engineering conventions: a
 repository whose files import `pyspark` has its Roles told to write DataFrame
 expressions rather than RDDs, and a Databricks workspace gets Unity Catalog
 naming and the Delta MERGE idioms in the prompts that write code and the
 workspace's own posture in the one that audits it. A repository that matches
-neither hears neither. `agentforge init` is still to come. See
-[`docs/PLAN.md`](docs/PLAN.md).
+neither hears neither. `agentforge init` inspects a repository and writes its
+`.agentforge/config.yaml`; the file is optional, and
+[Project configuration](#project-configuration) below says what it holds. What
+is still to come is in [`docs/PLAN.md`](docs/PLAN.md).
 
 Before the first Role is invoked, AgentForge resolves a Context Pack from the
 frozen plan — the files it names, the symbols and imports inside them, the
@@ -130,7 +132,7 @@ without notice. See
 ```console
 $ pip install agentforge-framework
 $ agentforge --version
-agentforge 0.2.4
+agentforge 0.2.5
 ```
 
 That is the whole of it. The two routes below the naming note are for a machine
@@ -154,7 +156,7 @@ from the [latest release](https://github.com/yashmhatre/AgentForge/releases/late
 and install the file:
 
 ```console
-$ pip install agentforge_framework-0.2.4-py3-none-any.whl
+$ pip install agentforge_framework-0.2.5-py3-none-any.whl
 ```
 
 Publishing a release is a wizard rather than a checklist:
@@ -191,7 +193,7 @@ AgentForge is a terminal tool. It runs fine from an IDE's terminal, but the IDE
 
 **Antigravity IDE cannot be a Provider** ([#101](https://github.com/yashmhatre/AgentForge/issues/101)). Checked by running it, so nobody has to derive it again: `antigravity-ide chat` accepts a prompt and reads stdin, which is the right shape for the argument vector — but it is a window launcher. Piping a prompt to it returns exit 0 immediately with `Reading from stdin via: …\code-stdin-XXXX` on stdout and nothing else, ever; the answer goes to a GUI session. A Provider adapter has to read the Agent's result off stdout, so there is nothing for `parse_output` to parse. The same is true of any editor CLI of this shape.
 
-**Do not run another agent against a checkout while a Run is going.** A Run commits every change to a file git already tracks, however it got there ([ADR-0015](docs/adr/0015-a-run-commits-what-it-declared.md)) — so a second agent's half-finished edits are committed into the Run's branch and attributed to a Role. Files nothing in the Run claimed are listed in the pull request body ([ADR-0023](docs/adr/0023-a-commit-names-what-nothing-claimed.md)), which makes it visible at Sign-off but does not prevent it. Use a separate clone.
+**Do not run another agent against a checkout while a Run is going.** A Run commits every change to a file git already tracks, however it got there ([ADR-0015](docs/adr/0015-a-run-commits-what-it-declared.md)) — so a second agent's half-finished edits are committed into the Run's branch and attributed to a Role. Files nothing in the Run claimed are listed in the pull request body ([ADR-0023](docs/adr/0023-a-commit-names-what-nothing-claimed.md)), which makes it visible at Sign-off but does not prevent it. Use a separate clone. A second `agentforge implement` is refused outright rather than merely disclosed — a Run holds its checkout for the duration ([ADR-0026](docs/adr/0026-a-run-holds-the-checkout-it-was-pointed-at.md)) — but nothing else consults that lock, so an IDE assistant is still yours to keep away.
 
 ## Commands
 
@@ -259,9 +261,21 @@ if the repository has no GitHub remote, because ADR-0002 makes that a
 precondition for every Run. Re-running never clobbers a config you have edited:
 it reports what differs and writes nothing, and `--force` replaces it.
 
+A Python suite is written pinned to the interpreter that will run it —
+`[".venv/bin/python", "-m", "pytest"]` rather than `pytest` — whenever the
+repository has a virtualenv to pin it to, and init checks that the interpreter
+can run pytest before writing the line. Bare `pytest` is whichever one PATH
+answers with, which in a project with a venv is the one the project does not
+use: the suite then fails on imports that are installed and reads as a broken
+repository rather than a misconfigured one. The interpreter is named relative to
+the repository root, so the file is still right on somebody else's clone, and
+init looks up to two directories down for the project — a suite that lives in
+`subproject/tests/` is found, and named.
+
 The file is not a precondition. Without one, the documented Provider capability
 defaults are Claude `native` and every other Provider `fragment`, and the
-`tests` Gate runs `pytest`.
+`tests` Gate runs `pytest` — the case the pinning above exists to avoid, so a
+repository with a venv is better off with a file.
 
 ```yaml
 providers:

{agentforge_framework-0.2.4 → agentforge_framework-0.2.5}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "agentforge-framework"
-version = "0.2.4"
+version = "0.2.5"
 description = "A modular framework for coordinating specialized software agents."
 readme = "README.md"
 requires-python = ">=3.11"

{agentforge_framework-0.2.4 → agentforge_framework-0.2.5}/src/agentforge_framework/__init__.py

@@ -1,3 +1,3 @@
 """AgentForge — coordination of specialized software agents through reusable workflows."""
 
-__version__ = "0.2.4"
+__version__ = "0.2.5"

{agentforge_framework-0.2.4 → agentforge_framework-0.2.5}/src/agentforge_framework/cli.py

@@ -512,7 +512,7 @@ def _run_init(args: argparse.Namespace, runner=None) -> int:
     from .core.config import load_config
     from .core.contracts import Plan
     from .core.process import SubprocessRunner
-    from .core.project import config_path, detect, differences, render_config
+    from .core.project import config_path, detect, differences, render_config, verified
     from .core.registry import activate
     from .core.repo import PreconditionFailed, open_repository
     from .providers import DEFAULT_PROVIDER, PROVIDERS
@@ -531,11 +531,14 @@ def _run_init(args: argparse.Namespace, runner=None) -> int:
         return 2
 
     active = activate(Plan(summary=""), repo.root)
-    context = detect(
-        repo.root,
-        provider,
-        tracked=repo.tracked_files(),
-        plugins=tuple(plugin.name for plugin in active.plugins),
+    context = verified(
+        detect(
+            repo.root,
+            provider,
+            tracked=repo.tracked_files(),
+            plugins=tuple(plugin.name for plugin in active.plugins),
+        ),
+        runner,
     )
 
     print(f"Repository: {repo.root}")
@@ -544,6 +547,8 @@ def _run_init(args: argparse.Namespace, runner=None) -> int:
     suite = " ".join(context.test_suite)
     where = context.suite_detected or "not detected, so this is the documented default"
     print(f"  Suite:     `{suite}` — {where}")
+    if context.suite_note:
+        print(f"             {context.suite_note}")
     print(f"  Plugins:   {', '.join(context.plugins) or 'none by root marker'}")
     print(
         "             printed, not written: which Plugins answer is decided per Run\n"

{agentforge_framework-0.2.4 → agentforge_framework-0.2.5}/src/agentforge_framework/core/gates.py

@@ -110,6 +110,11 @@ def human(context: GateContext) -> GateEntry:
 #: nothing collected; none of those say anything about the repository.
 SUITE_FAILED = 1
 
+#: What `python -m` prints when the runner it was asked for is not installed in
+#: that interpreter. It exits 1 doing it, which is the one status that otherwise
+#: means the suite ran and reported on the code.
+MODULE_MISSING = "No module named"
+
 #: How much of a failing suite reaches the Run Log. The end of it: a test runner
 #: puts its summary last, and a comment nobody scrolls to the bottom of is a
 #: comment nobody reads.
@@ -138,11 +143,12 @@ def tests(context: GateContext) -> GateEntry:
     runs. A Gate is not an Agent: the suite is the one the project declared, its
     exit status is read rather than interpreted, and no model chose either.
     """
-    suite = load_config(context.root).test_suite
-    rendered = " ".join(suite)
+    declared = load_config(context.root).test_suite
+    rendered = " ".join(declared)
+    suite = _resolved(declared, context.root)
 
     if not context.runner.has_binary(suite[0]):
-        return _cannot_run(rendered, f"{suite[0]!r} is not installed or not on PATH")
+        return _cannot_run(rendered, f"{declared[0]!r} is not installed or not on PATH")
 
     try:
         result = context.runner.run(suite, cwd=context.root)
@@ -160,6 +166,13 @@ def tests(context: GateContext) -> GateEntry:
             summary=f"`{rendered}` passed.",
         )
 
+    if result.returncode == SUITE_FAILED and MODULE_MISSING in result.stderr:
+        # An interpreter that has no pytest exits 1, which is the status a suite
+        # spends on real failures. The interpreter says so on standard error and
+        # a failing suite reports its failures on standard output, so the two are
+        # told apart there rather than by the status they share.
+        return _cannot_run(rendered, f"{declared[0]!r} has no test runner installed")
+
     if result.returncode == SUITE_FAILED:
         return GateEntry(
             kind="",
@@ -181,6 +194,25 @@ def tests(context: GateContext) -> GateEntry:
     )
 
 
+def _resolved(suite: tuple[str, ...], root: Path) -> tuple[str, ...]:
+    """A suite whose first word is a path, resolved against the repository.
+
+    `agentforge init` pins a Python suite to the project's own virtualenv and
+    writes that interpreter relative to the root, because the file it writes gets
+    committed and an absolute path is right on one machine only. Resolving it
+    here is what makes the relative form work at all: a relative program name is
+    resolved against the child process's directory on POSIX and the parent's on
+    Windows, so leaving it to the OS would run on one platform and not the other.
+
+    A suite naming a plain command is returned untouched, PATH lookup and all.
+    """
+    program = Path(suite[0])
+    if program.is_absolute() or len(program.parts) == 1:
+        return suite
+    candidate = root / program
+    return (str(candidate), *suite[1:]) if candidate.is_file() else suite
+
+
 def _cannot_run(rendered: str, reason: str) -> GateEntry:
     """The suite never started, which is not a report on the code.