aepkit 0.6.0__tar.gz → 0.7.0__tar.gz

{aepkit-0.6.0 → aepkit-0.7.0}/CHANGELOG.md

@@ -7,6 +7,30 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.7.0] - 2026-11-08
+
+### Added
+- `aepkit status --json` prints the status reading as one JSON object: each ID with its
+  phase, kind, effective state, verdict, latest report, executor (a status board column
+  named `Executor`, `Worker`, `Agent`, `Owner` or `Assignee`) and any board/header
+  conflict; the phases with their titles from `execution-guide.md` and the plan's phase
+  table, an optional `Weight` column, and phases the guide has not broken down; IDs the
+  guide lists that the log has no entry for yet, marked `inLog: false`. The text output
+  is unchanged.
+- `aeprogress`, an optional Claude Code mod in `integrations/claude-code/aeprogress/`:
+  a pane with the run's progress per phase, who is working on which ID (Claude Code
+  subagents, Orca workers, or any harness through a command), what awaits audit and what
+  needs attention, opened with `/aeprogress`. `.claude-plugin/marketplace.json` makes the
+  repository installable with
+  `/plugin install aeprogress --marketplace tBeltty/auditor-executor-protocol`.
+
+### Fixed
+- `SKILL.md`, `references/handoffs.md`, `references/autonomous-mode.md` and
+  `references/failure-modes-and-example.md` spoke of "a plain `PASS`" on one task. `PASS`
+  is not a verdict (the verdicts are `APPROVED`, `CONDITIONAL` and `REJECTED`), and a log
+  that used it showed the task as awaiting audit in `aepkit status`. The four sentences
+  now say `APPROVED`.
+
 ## [0.6.0] - 2026-10-24
 
 ### Added
@@ -29,6 +53,34 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   (`pipx install aepkit`, `aepkit`, `python -m aepkit`) to eliminate naming collisions with
   external `auditkit` packages and binaries.
 
+## [0.5.1] - 2026-10-04
+
+### Fixed
+- `negcontrol --file` without `--break-cmd` is now a usage error (exit 2). The backup was taken
+  of a file already broken by hand, so a correct `--restore-cmd` was overwritten with the broken
+  copy and blamed for it.
+- `negcontrol` fails the control when the test times out with the protection removed: a hang is
+  not a caught break. It used to print `OK` with a warning.
+- `negcontrol --file <directory>` and `auditkit init <existing file>` print an error and exit 2
+  instead of crashing with a traceback.
+- The execution guide's reporting format puts the verify output in a fenced block, so a report
+  written as the template shows no longer fails `lint`.
+- `lint` no longer cuts a report short at a `#`, `---` or `**Field:**` line inside pasted output.
+- `lint`'s phase-order check reads verdicts the way `status` does: the latest report header wins,
+  and a status board verdict that disagrees with it is a conflict, not an approval.
+- `install-skill` auto-detects Claude Code through `CLAUDECODE`, the variable Claude Code sets.
+- The install-skill tests no longer depend on the agent environment they run in.
+
+### Changed
+- The release workflow runs the full CI before building and publishing.
+- CI pins ruff, mypy, build and twine through the `dev` extra, and smoke-tests the built wheel in
+  a clean environment.
+- README links are absolute, so they work on PyPI; `auditkit --help` and the installed skill
+  link to GitHub instead of "the repo root".
+- `--annex-threshold` is documented as a total with a per-phase breakdown, which is what it
+  checks.
+- `SECURITY.md` lists 0.5.x as supported; `CONTRIBUTING.md` documents the release steps.
+
 ## [0.5.0] - 2026-10-02
 
 ### Added
@@ -85,9 +137,6 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   requires Python 3.11+.
 
 ### Fixed
-- `AGENTS.md` imported itself and said nothing, because `GEMINI.md` was a symlink to it. It is
-  now a short, generic agent guide that points to `CONTRIBUTING.md`; tool-specific instruction
-  files (`CLAUDE.md`, `GEMINI.md`) are no longer tracked.
 - README: the compliance log template, `install-skill` detection and `--global`/`--dest`, every
   command's options, exit codes, and the BSD-only `sed -i ''` in the `negcontrol` example now match
   the CLI.
@@ -193,7 +242,7 @@ Fixes from an independent review of 0.3.0.
   field is missing, empty, or only a template placeholder. The protocol already records
   these as `FAILED`; the check makes that mechanical.
 - README: install `auditkit` without cloning via `pipx install git+https://github.com/tBeltty/auditor-executor-protocol`.
-- Conventional Commits specification documented in `CONTRIBUTING.md` and enforced in `AGENTS.md`.
+- Conventional Commits specification documented in `CONTRIBUTING.md`.
 - `SKILL.md`: optional Guided/Autonomous operating mode, with the autonomy charter, per-task
   model-tier assignment, the per-task cycle, and an Auditor-brief template for starting an
   autonomous run.
@@ -218,8 +267,6 @@ Fixes from an independent review of 0.3.0.
 - `auditkit install-skill` copies `references/` beside `SKILL.md`. Destinations that hold a
   single file (Cursor `.mdc`, or a `--dest` not named `SKILL.md`) get one bundled file.
 - The packaged skill moved from `src/auditkit/templates/SKILL.md` to `src/auditkit/skill/`.
-- `GEMINI.md` imports `AGENTS.md` instead of duplicating it, so the agent rules have one
-  source of truth.
 - Bumped `actions/checkout` and `actions/setup-python` to v7 in GitHub Actions CI workflow.
 - Configured grouped updates in `.github/dependabot.yml` to bundle action and pip dependency bumps.
 
@@ -250,7 +297,6 @@ Fixes from an independent review of 0.3.0.
 - 90%+ test coverage enforcement across the repository.
 - GitHub Actions CI workflow hardening: explicit token permissions, dedicated lint/format/typecheck job, package distribution verification, and weekly Dependabot configuration.
 - Open-source governance files: `SECURITY.md`, `CONTRIBUTING.md`, `CHANGELOG.md`, PR and issue templates.
-- Agent rule specifications (`AGENTS.md`, `GEMINI.md`) establishing mandatory changelog updates and test verification gates on every task.
 
 ### Changed
 - Conformed all Python code to Ruff formatting and PEP 8 import conventions.
@@ -269,8 +315,10 @@ Fixes from an independent review of 0.3.0.
 - Multi-framework agent installation support for Antigravity, Claude Code, and Cursor.
 - Zero-dependency test runner (`tests/run.py`).
 
-[Unreleased]: https://github.com/tBeltty/auditor-executor-protocol/compare/v0.6.0...HEAD
-[0.6.0]: https://github.com/tBeltty/auditor-executor-protocol/compare/v0.5.0...v0.6.0
+[Unreleased]: https://github.com/tBeltty/auditor-executor-protocol/compare/v0.7.0...HEAD
+[0.7.0]: https://github.com/tBeltty/auditor-executor-protocol/compare/v0.6.0...v0.7.0
+[0.6.0]: https://github.com/tBeltty/auditor-executor-protocol/compare/v0.5.1...v0.6.0
+[0.5.1]: https://github.com/tBeltty/auditor-executor-protocol/compare/v0.5.0...v0.5.1
 [0.5.0]: https://github.com/tBeltty/auditor-executor-protocol/compare/v0.4.0...v0.5.0
 [0.4.0]: https://github.com/tBeltty/auditor-executor-protocol/compare/v0.3.9...v0.4.0
 [0.3.9]: https://github.com/tBeltty/auditor-executor-protocol/compare/v0.3.8...v0.3.9
@@ -284,3 +332,4 @@ Fixes from an independent review of 0.3.0.
 [0.3.1]: https://github.com/tBeltty/auditor-executor-protocol/compare/v0.3.0...v0.3.1
 [0.3.0]: https://github.com/tBeltty/auditor-executor-protocol/compare/v0.2.0...v0.3.0
 [0.2.0]: https://github.com/tBeltty/auditor-executor-protocol/releases/tag/v0.2.0
+[0.1.0]: https://github.com/tBeltty/auditor-executor-protocol/tree/7e9701cdacf732fa63b28428eedd9dcc7250fa2f

{aepkit-0.6.0/src/aepkit.egg-info → aepkit-0.7.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: aepkit
-Version: 0.6.0
+Version: 0.7.0
 Summary: Mechanical support for the Auditor/Executor protocol: scaffold, lint, and negative-control the document set for a multi-phase, multi-agent execution.
 Author-email: tBeltty <jhonatan@tbelt.online>
 License-Expression: MIT
@@ -30,9 +30,10 @@ Requires-Dist: pytest-cov>=4.0; extra == "test"
 Provides-Extra: dev
 Requires-Dist: pytest>=7.0; extra == "dev"
 Requires-Dist: pytest-cov>=4.0; extra == "dev"
-Requires-Dist: ruff>=0.1.0; extra == "dev"
-Requires-Dist: mypy>=1.0.0; extra == "dev"
-Requires-Dist: build>=1.0.0; extra == "dev"
+Requires-Dist: ruff==0.16.10; extra == "dev"
+Requires-Dist: mypy==2.4.0; extra == "dev"
+Requires-Dist: build==1.6.1; extra == "dev"
+Requires-Dist: twine==7.0.0; extra == "dev"
 Dynamic: license-file
 
 # Auditor/Executor Protocol
@@ -43,13 +44,13 @@ Dynamic: license-file
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
 [![Code style: ruff](https://img.shields.io/badge/code%20style-ruff-000000.svg)](https://github.com/astral-sh/ruff)
 [![Checked with mypy](https://img.shields.io/badge/mypy-strict-blue)](https://mypy-lang.org/)
-[![Zero Dependencies](https://img.shields.io/badge/dependencies-0-brightgreen)](pyproject.toml)
+[![Zero Dependencies](https://img.shields.io/badge/dependencies-0-brightgreen)](https://github.com/tBeltty/auditor-executor-protocol/blob/main/pyproject.toml)
 
 A two-role protocol for running multi-phase work through AI agents without the plan drifting into open-ended discussion and without "all tests pass" masquerading as verification.
 
 The **Auditor** defines what "done" means and proves it independently. The **Executor** implements one numbered task at a time and logs command output. Neither role crosses into the other.
 
-[`SKILL.md`](SKILL.md) specifies the protocol core; [`references/`](references/) holds the parts an agent loads only when it reaches that step (task and gate writing, handoff templates, Autonomous mode, failure modes and a worked example). `aepkit` is a zero-dependency Python CLI that automates scaffolding, drift linting, negative controls, status tallies, and installing the skill into your agent.
+[`SKILL.md`](https://github.com/tBeltty/auditor-executor-protocol/blob/main/SKILL.md) specifies the protocol core; [`references/`](https://github.com/tBeltty/auditor-executor-protocol/tree/main/references) holds the parts an agent loads only when it reaches that step (task and gate writing, handoff templates, Autonomous mode, failure modes and a worked example). `aepkit` is a zero-dependency Python CLI that automates scaffolding, drift linting, negative controls, status tallies, and installing the skill into your agent.
 
 ---
 
@@ -147,13 +148,15 @@ Inspect open tasks and tallied verdicts:
 aepkit status docs/<task-name>
 ```
 
+`aepkit status --json docs/<task-name>` prints the same reading as JSON, with each ID's phase, effective state and executor, and the phases with their titles from `execution-guide.md` (and weights from an optional `Weight` column in the plan's phase table), for scripts and tools to use instead of parsing the log.
+
 Cross-check document consistency:
 
 ```bash
 aepkit lint docs/<task-name>
 ```
 
-`aepkit lint` catches an empty guide, tasks without a report line, missing report entries, log entries with no matching task, `DONE` reports whose verify output is not pasted in a code block, gates without a stated negative control (hypothetical or waived ones do not count), work reported in a phase before every earlier phase is `APPROVED`, unfilled opportunity/shaping/non-goal fields in `plan-of-record.md` once work is reported `DONE` or audited, open rows in the deferred items ledger for a phase being closed, duplicated log paragraphs, and annex buildup past `--annex-threshold` (default 6, overall or per phase). It checks that evidence is present and in the right form, not that it is genuine; the Auditor's re-run does that.
+`aepkit lint` catches an empty guide, tasks without a report line, missing report entries, log entries with no matching task, `DONE` reports whose verify output is not pasted in a code block, gates without a stated negative control (hypothetical or waived ones do not count), work reported in a phase before every earlier phase is `APPROVED`, unfilled opportunity/shaping/non-goal fields in `plan-of-record.md` once work is reported `DONE` or audited, open rows in the deferred items ledger for a phase being closed, duplicated log paragraphs, and annex buildup past `--annex-threshold` (default 6 in total, with a per-phase breakdown). It checks that evidence is present and in the right form, not that it is genuine; the Auditor's re-run does that.
 
 ### 5. Run a Negative Control
 
@@ -169,9 +172,20 @@ aepkit negcontrol \
 
 The `sed -i ''` form above is BSD/macOS sed; on GNU/Linux use `sed -i 's/.../.../' file`.
 
-`negcontrol` backs up the target file, applies the break mutation, confirms the test fails, restores the file with byte-level verification, confirms the test passes, and outputs a transcript ready for `compliance-log.md`. Without `--file`, pass `--restore-cmd` to undo the break yourself; with both, the backup still wins if the file is not byte-identical afterwards. `--timeout` limits each command in seconds.
+`negcontrol` backs up the target file, applies the break mutation, confirms the test fails, restores the file with byte-level verification, confirms the test passes, and outputs a transcript ready for `compliance-log.md`. Without `--file`, pass `--restore-cmd` to undo the break yourself; with both, the backup still wins if the file is not byte-identical afterwards. `--file` needs `--break-cmd`, because the backup must be taken before the break. `--timeout` limits each command in seconds, and a test that times out instead of failing fails the control: a hang is not a caught break.
+
+A failure only proves something if it is the failure you meant. A syntax error, a missing import or a mistyped test command also makes the test exit non-zero. `--expect` takes a regular expression that the failing run's output must match, and the control fails if it does not. Without `--expect`, output that looks like a syntax, import, collection or command-not-found error is flagged with a `WARNING` next to the `OK`, and you need to read the transcript to confirm the reason.
+
+### Optional: Progress Pane for Claude Code
+
+`aeprogress` is a Claude Code mod that keeps a pane beside the chat with the run's overall percentage, a bar per phase, who is working on which task ID (Claude Code subagents, Orca workers, or any other harness through a command of your own), what awaits audit and what needs attention. It reads the log through `aepkit status --json` and refreshes after each turn, and every 20 seconds while the pane is open. The protocol does not depend on it.
+
+```
+/plugin install aeprogress --marketplace tBeltty/auditor-executor-protocol
+/aeprogress docs/<task-name>
+```
 
-A failure only proves something if it is the failure you meant. A syntax error, a missing import or a mistyped test command also makes the test exit non-zero. `--expect` takes a regular expression that the failing run's output must match, and the control fails if it does not. Without `--expect`, output that looks like a syntax, import, collection, timeout or command-not-found error is flagged with a `WARNING` next to the `OK`, and you need to read the transcript to confirm the reason.
+It works in the terminal and in the desktop app's local sessions. Claude Code's mod API is early access, so a Claude Code release can break it. Setup, options and how the percentage is computed: [integrations/claude-code/aeprogress](https://github.com/tBeltty/auditor-executor-protocol/tree/main/integrations/claude-code/aeprogress).
 
 ---
 
@@ -183,7 +197,7 @@ A failure only proves something if it is the failure you meant. A syntax error,
 | `aepkit install-skill [dir]` | `--agent`, `--global`, `--dest`, `--force` | Provision `SKILL.md` and `references/` into Antigravity, Claude Code, or Cursor |
 | `aepkit lint <dir>` | `--annex-threshold` | Cross-check IDs both ways, flag `DONE` without evidence, detect missing negative controls, and flag log rot |
 | `aepkit negcontrol` | `--test-cmd` (required), `--file`, `--break-cmd`, `--restore-cmd`, `--expect`, `--timeout` | Run automated backup, break, fail, restore, and pass cycle |
-| `aepkit status <dir>` | | Combine status board verdicts with reports; list everything not `APPROVED` |
+| `aepkit status <dir>` | `--json` | Combine status board verdicts with reports; list everything not `APPROVED` |
 
 `python -m aepkit` works the same as `aepkit`. Errors go to stderr.
 
@@ -212,11 +226,11 @@ pytest
 
 ## Contributing
 
-Contributions are welcome. [`CONTRIBUTING.md`](CONTRIBUTING.md) covers the development setup, the quality gates, and the PR checklist; participation follows the [`CODE_OF_CONDUCT.md`](CODE_OF_CONDUCT.md). Report vulnerabilities privately as described in [`SECURITY.md`](SECURITY.md). Changes are listed in [`CHANGELOG.md`](CHANGELOG.md).
+Contributions are welcome. [`CONTRIBUTING.md`](https://github.com/tBeltty/auditor-executor-protocol/blob/main/CONTRIBUTING.md) covers the development setup, the quality gates, and the PR checklist; participation follows the [`CODE_OF_CONDUCT.md`](https://github.com/tBeltty/auditor-executor-protocol/blob/main/CODE_OF_CONDUCT.md). Report vulnerabilities privately as described in [`SECURITY.md`](https://github.com/tBeltty/auditor-executor-protocol/blob/main/SECURITY.md). Changes are listed in [`CHANGELOG.md`](https://github.com/tBeltty/auditor-executor-protocol/blob/main/CHANGELOG.md).
 
 ## License
 
-MIT (see [`LICENSE`](LICENSE)).
+MIT (see [`LICENSE`](https://github.com/tBeltty/auditor-executor-protocol/blob/main/LICENSE)).
 
 ---
 

{aepkit-0.6.0 → aepkit-0.7.0}/README.md

@@ -6,13 +6,13 @@
 [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
 [![Code style: ruff](https://img.shields.io/badge/code%20style-ruff-000000.svg)](https://github.com/astral-sh/ruff)
 [![Checked with mypy](https://img.shields.io/badge/mypy-strict-blue)](https://mypy-lang.org/)
-[![Zero Dependencies](https://img.shields.io/badge/dependencies-0-brightgreen)](pyproject.toml)
+[![Zero Dependencies](https://img.shields.io/badge/dependencies-0-brightgreen)](https://github.com/tBeltty/auditor-executor-protocol/blob/main/pyproject.toml)
 
 A two-role protocol for running multi-phase work through AI agents without the plan drifting into open-ended discussion and without "all tests pass" masquerading as verification.
 
 The **Auditor** defines what "done" means and proves it independently. The **Executor** implements one numbered task at a time and logs command output. Neither role crosses into the other.
 
-[`SKILL.md`](SKILL.md) specifies the protocol core; [`references/`](references/) holds the parts an agent loads only when it reaches that step (task and gate writing, handoff templates, Autonomous mode, failure modes and a worked example). `aepkit` is a zero-dependency Python CLI that automates scaffolding, drift linting, negative controls, status tallies, and installing the skill into your agent.
+[`SKILL.md`](https://github.com/tBeltty/auditor-executor-protocol/blob/main/SKILL.md) specifies the protocol core; [`references/`](https://github.com/tBeltty/auditor-executor-protocol/tree/main/references) holds the parts an agent loads only when it reaches that step (task and gate writing, handoff templates, Autonomous mode, failure modes and a worked example). `aepkit` is a zero-dependency Python CLI that automates scaffolding, drift linting, negative controls, status tallies, and installing the skill into your agent.
 
 ---
 
@@ -110,13 +110,15 @@ Inspect open tasks and tallied verdicts:
 aepkit status docs/<task-name>
 ```
 
+`aepkit status --json docs/<task-name>` prints the same reading as JSON, with each ID's phase, effective state and executor, and the phases with their titles from `execution-guide.md` (and weights from an optional `Weight` column in the plan's phase table), for scripts and tools to use instead of parsing the log.
+
 Cross-check document consistency:
 
 ```bash
 aepkit lint docs/<task-name>
 ```
 
-`aepkit lint` catches an empty guide, tasks without a report line, missing report entries, log entries with no matching task, `DONE` reports whose verify output is not pasted in a code block, gates without a stated negative control (hypothetical or waived ones do not count), work reported in a phase before every earlier phase is `APPROVED`, unfilled opportunity/shaping/non-goal fields in `plan-of-record.md` once work is reported `DONE` or audited, open rows in the deferred items ledger for a phase being closed, duplicated log paragraphs, and annex buildup past `--annex-threshold` (default 6, overall or per phase). It checks that evidence is present and in the right form, not that it is genuine; the Auditor's re-run does that.
+`aepkit lint` catches an empty guide, tasks without a report line, missing report entries, log entries with no matching task, `DONE` reports whose verify output is not pasted in a code block, gates without a stated negative control (hypothetical or waived ones do not count), work reported in a phase before every earlier phase is `APPROVED`, unfilled opportunity/shaping/non-goal fields in `plan-of-record.md` once work is reported `DONE` or audited, open rows in the deferred items ledger for a phase being closed, duplicated log paragraphs, and annex buildup past `--annex-threshold` (default 6 in total, with a per-phase breakdown). It checks that evidence is present and in the right form, not that it is genuine; the Auditor's re-run does that.
 
 ### 5. Run a Negative Control
 
@@ -132,9 +134,20 @@ aepkit negcontrol \
 
 The `sed -i ''` form above is BSD/macOS sed; on GNU/Linux use `sed -i 's/.../.../' file`.
 
-`negcontrol` backs up the target file, applies the break mutation, confirms the test fails, restores the file with byte-level verification, confirms the test passes, and outputs a transcript ready for `compliance-log.md`. Without `--file`, pass `--restore-cmd` to undo the break yourself; with both, the backup still wins if the file is not byte-identical afterwards. `--timeout` limits each command in seconds.
+`negcontrol` backs up the target file, applies the break mutation, confirms the test fails, restores the file with byte-level verification, confirms the test passes, and outputs a transcript ready for `compliance-log.md`. Without `--file`, pass `--restore-cmd` to undo the break yourself; with both, the backup still wins if the file is not byte-identical afterwards. `--file` needs `--break-cmd`, because the backup must be taken before the break. `--timeout` limits each command in seconds, and a test that times out instead of failing fails the control: a hang is not a caught break.
+
+A failure only proves something if it is the failure you meant. A syntax error, a missing import or a mistyped test command also makes the test exit non-zero. `--expect` takes a regular expression that the failing run's output must match, and the control fails if it does not. Without `--expect`, output that looks like a syntax, import, collection or command-not-found error is flagged with a `WARNING` next to the `OK`, and you need to read the transcript to confirm the reason.
+
+### Optional: Progress Pane for Claude Code
+
+`aeprogress` is a Claude Code mod that keeps a pane beside the chat with the run's overall percentage, a bar per phase, who is working on which task ID (Claude Code subagents, Orca workers, or any other harness through a command of your own), what awaits audit and what needs attention. It reads the log through `aepkit status --json` and refreshes after each turn, and every 20 seconds while the pane is open. The protocol does not depend on it.
+
+```
+/plugin install aeprogress --marketplace tBeltty/auditor-executor-protocol
+/aeprogress docs/<task-name>
+```
 
-A failure only proves something if it is the failure you meant. A syntax error, a missing import or a mistyped test command also makes the test exit non-zero. `--expect` takes a regular expression that the failing run's output must match, and the control fails if it does not. Without `--expect`, output that looks like a syntax, import, collection, timeout or command-not-found error is flagged with a `WARNING` next to the `OK`, and you need to read the transcript to confirm the reason.
+It works in the terminal and in the desktop app's local sessions. Claude Code's mod API is early access, so a Claude Code release can break it. Setup, options and how the percentage is computed: [integrations/claude-code/aeprogress](https://github.com/tBeltty/auditor-executor-protocol/tree/main/integrations/claude-code/aeprogress).
 
 ---
 
@@ -146,7 +159,7 @@ A failure only proves something if it is the failure you meant. A syntax error,
 | `aepkit install-skill [dir]` | `--agent`, `--global`, `--dest`, `--force` | Provision `SKILL.md` and `references/` into Antigravity, Claude Code, or Cursor |
 | `aepkit lint <dir>` | `--annex-threshold` | Cross-check IDs both ways, flag `DONE` without evidence, detect missing negative controls, and flag log rot |
 | `aepkit negcontrol` | `--test-cmd` (required), `--file`, `--break-cmd`, `--restore-cmd`, `--expect`, `--timeout` | Run automated backup, break, fail, restore, and pass cycle |
-| `aepkit status <dir>` | | Combine status board verdicts with reports; list everything not `APPROVED` |
+| `aepkit status <dir>` | `--json` | Combine status board verdicts with reports; list everything not `APPROVED` |
 
 `python -m aepkit` works the same as `aepkit`. Errors go to stderr.
 
@@ -175,11 +188,11 @@ pytest
 
 ## Contributing
 
-Contributions are welcome. [`CONTRIBUTING.md`](CONTRIBUTING.md) covers the development setup, the quality gates, and the PR checklist; participation follows the [`CODE_OF_CONDUCT.md`](CODE_OF_CONDUCT.md). Report vulnerabilities privately as described in [`SECURITY.md`](SECURITY.md). Changes are listed in [`CHANGELOG.md`](CHANGELOG.md).
+Contributions are welcome. [`CONTRIBUTING.md`](https://github.com/tBeltty/auditor-executor-protocol/blob/main/CONTRIBUTING.md) covers the development setup, the quality gates, and the PR checklist; participation follows the [`CODE_OF_CONDUCT.md`](https://github.com/tBeltty/auditor-executor-protocol/blob/main/CODE_OF_CONDUCT.md). Report vulnerabilities privately as described in [`SECURITY.md`](https://github.com/tBeltty/auditor-executor-protocol/blob/main/SECURITY.md). Changes are listed in [`CHANGELOG.md`](https://github.com/tBeltty/auditor-executor-protocol/blob/main/CHANGELOG.md).
 
 ## License
 
-MIT (see [`LICENSE`](LICENSE)).
+MIT (see [`LICENSE`](https://github.com/tBeltty/auditor-executor-protocol/blob/main/LICENSE)).
 
 ---
 

{aepkit-0.6.0 → aepkit-0.7.0}/SKILL.md

@@ -188,7 +188,6 @@ done before task 1 is drafted:**
   | Rule | Look for |
   |---|---|
   | Version bump per change | A version field (`package.json`, `pyproject.toml`, a `VERSION` file) and any script that already bumps it |
-  | No AI co-author trailers | `git log --grep="Co-Authored-By"` — existing trailers (or their consistent absence) |
   | Fetch-before-push | Multiple contributors or branches in `git log`/`git branch -a` — a solo, single-branch repo needs this less |
   | Quality gate green before push | CI config (`.github/workflows/`, `.gitlab-ci.yml`) or a `test`/`lint`/`build` script already wired together |
   | i18n/locale parity | A locales/translations directory with more than one language file |
@@ -299,7 +298,7 @@ report's job is to point at the evidence, not to argue for a verdict.
    for the whole feature.
 8. **Before sending the verdict, check what it's missing, not just what it says.** This
    applies to every verdict, not only a full-phase `APPROVED` or a `CONDITIONAL`/
-   `REJECTED` with an obvious remediation — a plain `PASS` on one task, mid-phase, with
+   `REJECTED` with an obvious remediation — a plain `APPROVED` on one task, mid-phase, with
    something else already outstanding (an earlier remediation, the next task) needs the
    same discipline: it does not end the message on its own. If anything is left to do —
    a remediation handoff, the next task's handoff, or a re-statement of a handoff already
@@ -423,7 +422,7 @@ If `aepkit` is installed, prefer it over doing these by hand:
   compliance log in both directions, flag `DONE` reports with no pasted verify output
   (output counts only inside a fenced code block; prose such as "all tests pass" or a lone
   "n/a" is not output), flag work reported in a phase before every earlier phase is
-  `APPROVED`, flag near-duplicate paragraphs, warn on annex count per phase, flag gates with no
+  `APPROVED`, flag near-duplicate paragraphs, warn on annex count (listed per phase), flag gates with no
   stated negative control (it must name the action that removes the protection and the
   failure that follows; a negated, deferred, hypothetical, or waived mention such as "n/a",
   "TBD", or "we could remove it" does not count), and—once work has started (`DONE` or a
@@ -437,12 +436,19 @@ If `aepkit` is installed, prefer it over doing these by hand:
   file up, runs the break command, runs the test (expects failure), restores from the
   backup, runs the test again (expects success), and prints a paste-ready transcript.
   With `--restore-cmd`, the file must still end byte-identical to the backup or it is
-  restored from it; `--timeout` bounds each command. Pass `--expect "<regex>"` with the
+  restored from it. `--file` needs `--break-cmd`, since the backup must come before the
+  break. `--timeout` bounds each command, and a test that times out instead of failing
+  fails the control. Pass `--expect "<regex>"` with the
   failure the break should cause (an assertion message, a 403): a test that fails for
   another reason, such as a syntax error or a missing import, then fails the control
   instead of passing it. Without `--expect`, such output is only flagged as a warning.
 - `aepkit status <dir>` — combines the status board verdicts with the latest report
   per ID and prints what's still open. Only `APPROVED` closes an item; a `DONE` with no
-  verdict is listed as awaiting audit.
-
-See the repo's `README.md` for install instructions.
+  verdict is listed as awaiting audit. `--json` prints the same reading, with the phases
+  from the guide and the plan, for tools to use instead of parsing the log themselves.
+- Optional, Claude Code only: the `aeprogress` mod keeps a pane beside the chat with the
+  run's progress per phase and who is working on which ID (subagents, Orca workers, or
+  any harness through a command). The protocol does not depend on it. See
+  `integrations/claude-code/aeprogress/` in the repository.
+
+Install instructions: https://github.com/tBeltty/auditor-executor-protocol#quickstart

{aepkit-0.6.0 → aepkit-0.7.0}/pyproject.toml

@@ -45,12 +45,15 @@ test = [
     "pytest>=7.0",
     "pytest-cov>=4.0",
 ]
+# Exact pins, so a new linter rule or build-tool release cannot turn CI red or change a
+# published build without a commit; Dependabot bumps them.
 dev = [
     "pytest>=7.0",
     "pytest-cov>=4.0",
-    "ruff>=0.1.0",
-    "mypy>=1.0.0",
-    "build>=1.0.0",
+    "ruff==0.16.10",
+    "mypy==2.4.0",
+    "build==1.6.1",
+    "twine==7.0.0",
 ]
 
 [project.scripts]
@@ -68,6 +71,8 @@ where = ["src"]
 [tool.ruff]
 line-length = 100
 target-version = "py311"
+# agent-produced run directories and experiment scripts are data, not package code
+extend-exclude = ["benchmarks"]
 
 [tool.ruff.lint]
 select = ["E", "F", "W", "I", "UP", "B", "PL", "RUF"]
@@ -82,7 +87,7 @@ ignore = [
 ]
 
 [tool.ruff.lint.per-file-ignores]
-"tests/*" = ["RUF001", "BLE001"]
+"tests/*" = ["RUF001"]
 
 [tool.mypy]
 python_version = "3.11"

{aepkit-0.6.0/src/aepkit/skill → aepkit-0.7.0}/references/autonomous-mode.md

@@ -47,6 +47,10 @@ it from a governance decision in the plan of record. It has four parts:
    made: <decision-entry IDs, or "none">. Next: <phase/task already launched>.
    ```
 
+   Between reports, an owner on Claude Code can follow the run in the optional
+   `aeprogress` pane (`integrations/claude-code/aeprogress/` in the repository), which
+   reads the compliance log and never messages the Auditor. The run does not depend on it.
+
    Harness permission prompts that need the owner go in **one batched request**, not one
    at a time. When every phase is `APPROVED` and the plan of record declares no further
    phase, say so as its own line — see "Closing the run" in `SKILL.md`.
@@ -91,7 +95,7 @@ each handoff. The shape that has worked here:
    its first task.
 
 Do not end a turn between a verdict and the next launch unless something on the
-critical-risk list is open. A run that stops to report a `PASS` and wait is Guided
+critical-risk list is open. A run that stops to report an `APPROVED` and wait is Guided
 mode by accident.
 
 ### Starting the Auditor (the Auditor brief)

{aepkit-0.6.0/src/aepkit/skill → aepkit-0.7.0}/references/failure-modes-and-example.md

@@ -23,7 +23,7 @@ this file.
 - **Silent model inheritance.** A subagent launched without an explicit model runs on
   the Auditor's own model. Nothing fails; the rule is just broken on every task. Pass
   it every time.
-- **Guided by habit.** In Autonomous mode, the Auditor ends a turn after a `PASS` to
+- **Guided by habit.** In Autonomous mode, the Auditor ends a turn after an `APPROVED` to
   "report," or asks the owner a question the charter already lets it decide. Each one
   costs the owner a message and adds nothing.
 - **A defect ships and nobody owns re-checking its blast radius.** A task changes what

{aepkit-0.6.0/src/aepkit/skill → aepkit-0.7.0}/references/handoffs.md

@@ -120,7 +120,7 @@ Auditor's own rule 4 ("The Auditor is bound by rule 4 too" in `SKILL.md`).
 
 **No verdict ends the loop by itself if anything is left to run — and "anything left to
 run" is not only "a phase remains" or "a remediation order is needed."** It also covers
-the most common case of all: a plain `PASS` on one task, mid-phase, with something else
+the most common case of all: a plain `APPROVED` on one task, mid-phase, with something else
 already outstanding — an earlier remediation, the next task in line. That case has no
 name of its own in the Verdicts table of `SKILL.md`, which is exactly why it's the easiest one
 to ship without a handoff: closing with a status sentence that *describes* what's still

aepkit-0.7.0/src/aepkit/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.7.0"

{aepkit-0.6.0 → aepkit-0.7.0}/src/aepkit/cli.py

@@ -1,9 +1,11 @@
 """aepkit — mechanical support for the Auditor/Executor protocol.
 
-See SKILL.md in the repo root for the protocol itself. This CLI operationalizes the
-parts of it that are checklist work, not judgment: scaffolding the document set,
-cross-checking it for drift, running a negative control end to end, and reporting
-status.
+The protocol itself is SKILL.md:
+https://github.com/tBeltty/auditor-executor-protocol/blob/main/SKILL.md
+
+This CLI operationalizes the parts of it that are checklist work, not judgment:
+scaffolding the document set, cross-checking it for drift, running a negative control
+end to end, and reporting status.
 """
 
 from __future__ import annotations
@@ -15,7 +17,9 @@ from . import __version__, install_skill, lint, negcontrol, scaffold, status
 
 
 def build_parser() -> argparse.ArgumentParser:
-    parser = argparse.ArgumentParser(prog="aepkit", description=__doc__)
+    parser = argparse.ArgumentParser(
+        prog="aepkit", description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter
+    )
     parser.add_argument("--version", action="version", version=f"aepkit {__version__}")
     sub = parser.add_subparsers(dest="command", required=True)
 
@@ -39,7 +43,7 @@ def build_parser() -> argparse.ArgumentParser:
         type=int,
         metavar="N",
         default=lint.DEFAULT_ANNEX_THRESHOLD,
-        help="flag more annexes than this, overall or per phase (default: %(default)s)",
+        help="flag more annexes than this in total, listed per phase (default: %(default)s)",
     )
 
     p_neg = sub.add_parser("negcontrol", help="run a backup/break/test/restore/test cycle")
@@ -85,6 +89,12 @@ def build_parser() -> argparse.ArgumentParser:
         "status", help="tally verdicts and reports in the compliance log; only APPROVED closes"
     )
     p_status.add_argument("target_dir", help="directory holding compliance-log.md")
+    p_status.add_argument(
+        "--json",
+        dest="as_json",
+        action="store_true",
+        help="print the reading as JSON, with phases from the execution guide and the plan",
+    )
 
     p_install = sub.add_parser(
         "install-skill", help="provision SKILL.md and references/ into an agent skills directory"
@@ -137,7 +147,7 @@ def main(argv: list[str] | None = None) -> int:
             expect=args.expect,
         )
     if args.command == "status":
-        return status.run(args.target_dir)
+        return status.run(args.target_dir, as_json=args.as_json)
     return install_skill.run(
         target_dir=args.target_dir,
         agent=args.agent,

{aepkit-0.6.0 → aepkit-0.7.0}/src/aepkit/install_skill.py

@@ -61,7 +61,7 @@ def _detect_agent(target: Path) -> str:
     # 2. Check runtime environment variables
     if os.environ.get("ANTIGRAVITY_AGENT") or os.environ.get("GEMINI_CLI"):
         return "antigravity"
-    if os.environ.get("CLAUDE_CODE") or os.environ.get("CLAUDE_PROJECT_DIR"):
+    if any(os.environ.get(v) for v in ("CLAUDECODE", "CLAUDE_CODE", "CLAUDE_PROJECT_DIR")):
         return "claude"
     if os.environ.get("CURSOR_PROJECT_DIR"):
         return "cursor"

{aepkit-0.6.0 → aepkit-0.7.0}/src/aepkit/lint.py

@@ -13,6 +13,8 @@ import sys
 from collections import Counter, defaultdict
 from pathlib import Path
 
+from .status import VERDICTS, _cells, board_verdicts
+
 REPORT_ID_RE = re.compile(r"\*\*Report:\*\*\s*`?([A-Za-z0-9_.\-]+)`?")
 GUIDE_HEADER_RE = re.compile(r"^###\s+([A-Za-z0-9_.\-]+)\s+[\u2014\u2013\-]\s*(.*)$", re.MULTILINE)
 LOG_HEADER_RE = re.compile(
@@ -95,7 +97,6 @@ FENCE_LINE_RE = re.compile(r"^\s*(```|~~~)[\w-]*\s*$", re.MULTILINE)
 FENCED_BLOCK_RE = re.compile(r"^\s*(```|~~~)[\w-]*\s*\n(.*?)^\s*\1\s*$", re.MULTILINE | re.DOTALL)
 PHASE_RE = re.compile(r"^P(\d+)-", re.IGNORECASE)
 DEFERRED_PHASE_RE = re.compile(r"\b(?:P|Phase\s*)(\d+)\b", re.IGNORECASE)
-VERDICTS = {"APPROVED", "CONDITIONAL", "REJECTED"}
 DETERMINISTIC_BYPASS_RE = re.compile(
     r"\b(?:n/?a|not\s+applicable)\b[^\n]{0,60}\b"
     r"(?:deterministic|bug\s*fix|fix|compliance|security|patch|migration)\b",
@@ -170,6 +171,11 @@ def _is_gate(task_id: str, title: str) -> bool:
     return bool(GATE_ID_RE.search(task_id)) or "gate" in title.lower()
 
 
+def _mask_fences(text: str) -> str:
+    """text with every fenced block blanked out, keeping offsets and line breaks."""
+    return FENCED_BLOCK_RE.sub(lambda b: re.sub(r"[^\n]", " ", b.group(0)), text)
+
+
 def _done_without_evidence(log_text: str) -> list[str]:
     """IDs reported DONE whose **Verify output:** field is missing, empty, or only defers.
 
@@ -183,15 +189,16 @@ def _done_without_evidence(log_text: str) -> list[str]:
         if m.group(2).upper() != "DONE":
             continue
         rest = log_text[m.end() :]
-        end = SECTION_END_RE.search(rest)
-        body = rest[: end.start()] if end else rest
-        field = VERIFY_FIELD_RE.search(body)
+        # A "#", "---" or "**Field:**" line inside pasted output does not end the report.
+        masked = _mask_fences(rest)
+        end = SECTION_END_RE.search(masked)
+        body_end = end.start() if end else len(rest)
+        field = VERIFY_FIELD_RE.search(masked, 0, body_end)
         if not field:
             missing.append(m.group(1))
             continue
-        after = body[field.end() :]
-        nxt = NEXT_FIELD_RE.search(after)
-        evidence = after[: nxt.start()] if nxt else after
+        nxt = NEXT_FIELD_RE.search(masked, field.end(), body_end)
+        evidence = rest[field.end() : nxt.start() if nxt else body_end]
         blocks = [
             PLACEHOLDER_RE.sub("", b.group(2)).strip() for b in FENCED_BLOCK_RE.finditer(evidence)
         ]
@@ -202,18 +209,23 @@ def _done_without_evidence(log_text: str) -> list[str]:
 
 def _phase_order_violations(guide_text: str, log_text: str) -> list[str]:
     """IDs reported DONE in a phase that started before every earlier phase was APPROVED."""
-    from .status import _board_verdicts
-
     ids = [task_id for task_id, _, _ in _guide_sections(guide_text) if TASK_ID_RE.search(task_id)]
-    verdicts = {k: v for k, v in _board_verdicts(log_text).items() if v in VERDICTS}
+    board = {k: v for k, v in board_verdicts(log_text).items() if v in VERDICTS}
+    # As in `aepkit status`: the last report header per ID wins, and a board verdict that
+    # disagrees with the header verdict is a conflict, which does not count as APPROVED.
+    header_verdicts: dict[str, str] = {}
     reports: dict[str, str] = {}
     for m in LOG_HEADER_RE.finditer(log_text):
         task_id, value = m.group(1), m.group(2).upper()
         ids.append(task_id)
         if value in VERDICTS:
-            verdicts.setdefault(task_id, value)
+            header_verdicts[task_id] = value
         else:
             reports[task_id] = value
+    verdicts = {**header_verdicts, **board}
+    for task_id, value in header_verdicts.items():
+        if board.get(task_id, value) != value:
+            verdicts[task_id] = "CONFLICT"
     phases: defaultdict[int, set[str]] = defaultdict(set)
     for task_id in ids:
         found = PHASE_RE.match(task_id)
@@ -291,9 +303,7 @@ def _plan_gaps(plan_text: str) -> list[str]:
 def _open_deferred_items(plan_text: str, known_ids: set[str], log_text: str) -> list[str]:
     """Flag deferred-ledger rows whose closing task ID does not exist or remains empty when
     its target phase is being closed."""
-    from .status import _board_verdicts, _cells
-
-    verdicts = {k: v for k, v in _board_verdicts(log_text).items() if v in VERDICTS}
+    verdicts = {k: v for k, v in board_verdicts(log_text).items() if v in VERDICTS}
     reports: dict[str, str] = {}
     for m in LOG_HEADER_RE.finditer(log_text):
         tid, val = m.group(1), m.group(2).upper()
@@ -476,10 +486,8 @@ def run(target_dir: str, annex_threshold: int = DEFAULT_ANNEX_THRESHOLD) -> int:
     # 3b. Plan of record: opportunity/shaping/non-goals (once work starts) and deferred items.
     plan_path = target / "plan-of-record.md"
     if plan_path.is_file():
-        from .status import _board_verdicts
-
         plan_text = _read(plan_path)
-        board = _board_verdicts(log_text)
+        board = board_verdicts(log_text)
         work_started = any(
             v == "DONE" or v in VERDICTS for v in (*log_status.values(), *board.values())
         )
@@ -528,8 +536,7 @@ def run(target_dir: str, annex_threshold: int = DEFAULT_ANNEX_THRESHOLD) -> int:
             else:
                 phase_counts["General"] += 1
 
-        phase_exceeded = {p: c for p, c in phase_counts.items() if c > annex_threshold}
-        if len(annex_files) > annex_threshold or phase_exceeded:
+        if len(annex_files) > annex_threshold:
             problems += 1
             print(
                 f"[annex count] {len(annex_files)} annexes exceeds the threshold of "