[feat][#16] Re-implement mega-planner as standalone Python script by Tonny-Gu · Pull Request #17 · Tonny-Gu/agentize

Tonny-Gu · 2026-02-08T12:04:30Z

[feat][#16] Re-implement mega-planner as standalone Python script

Summary

Re-implemented the mega-planner (previously a Claude Code plugin command in .claude-plugin/commands/mega-planner.md) as a standalone Python script. The script orchestrates a 7-stage multi-agent debate pipeline using the agentize.workflow.api Session DSL, with co-located prompt files and full 4-mode CLI support.

Changes

Created scripts/mega-planner.py (~500 LOC) implementing the full 7-stage pipeline: understander -> (bold + paranoia in parallel) -> (critique + proposal-reducer + code-reducer in parallel) -> consensus
Supports 4 CLI modes: default (--feature-desc), --from-issue, --refine-issue, --resolve-issue
Copied 7 agent prompt files to scripts/prompts/ for co-location with the script
Created scripts/prompts/README.md documenting prompt file sources and synchronization
Created scripts/mega-planner.md documenting script interfaces, CLI modes, and internal helpers
Updated scripts/README.md with mega-planner section
Added python/tests/test_mega_planner.py with 12 tests covering pipeline stages, execution order, prompt rendering, resolve mode, skip_consensus, and feature name extraction
Deleted deprecated shim python/agentize/workflow/planner.py (dead code, shadowed by planner/ package directory, marked "TODO: Delete")
Deleted python/agentize/workflow/planner.md (docs for deprecated shim)

Testing

Added python/tests/test_mega_planner.py with 12 tests verifying:
- All 7 stages return StageResult (test_returns_all_seven_stages)
- skip_consensus returns 6 stages (test_skip_consensus)
- Resolve mode skips debate and runs consensus only (test_resolve_mode_skips_debate)
- Output artifacts created for each stage (test_output_artifacts_created)
- Debate report saved with correct sections (test_debate_report_saved)
- Understander runs before proposers (test_understander_runs_before_proposers)
- Bold + paranoia dispatched in parallel (test_bold_paranoia_parallel)
- Feature description appears in rendered prompts (test_feature_description_in_prompts)
- Dual-input stages receive both proposals (test_dual_input_stages_have_both_proposals)
- Feature name extraction: short, long truncation, multiline (3 tests)
All 277 project tests pass (12 new mega-planner + 22 existing planner + 243 other)

Related Issue

Closes #16

Add `make env` and `make env-script` targets for Makefile-centric environment setup as alternative to `source setup.sh`. - Makefile: add `env` target (exports AGENTIZE_HOME, PYTHONPATH) and document it in `make help` output. - templates/python/Makefile: add `env` (PROJECT_ROOT, PATH, PYTHONPATH), `env-script` (generates setup.sh), and `help` targets. - templates/c/Makefile: add `env` (PROJECT_ROOT, PATH, C_INCLUDE_PATH, LIBRARY_PATH), `env-script`, and `help` targets. - templates/cxx/Makefile: add `env` (PROJECT_ROOT, PATH, CPLUS_INCLUDE_PATH, LIBRARY_PATH), `env-script`, and `help` targets. - templates/README.md: document new env, env-script, and help targets. - tests/lint/test-makefile-env-target.sh: verify make env output and eval correctness. - tests/lint/test-makefile-help-env-targets.sh: verify make help documents env target. - tests/sdk/test-template-env-targets.sh: verify all template Makefiles have env, env-script, help targets with language-specific exports. Test status: 60/60 shell tests passed. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Copilot

Pull request overview

Adds make env / make env-script / make help targets to the main Makefile and SDK templates to standardize environment setup (either via eval $(make env) or generating setup.sh), and introduces shell tests to validate these targets.

Changes:

Add env, env-script, and help targets to the Python/C/C++ template Makefiles with language-specific exports.
Add env target (and help text) to the repo root Makefile.
Add lint/sdk shell tests to verify make env output and template target presence.

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
`Makefile`	Adds `env` target and documents it in `make help`.
`templates/python/Makefile`	Adds `env`, `env-script`, `help` targets for Python SDK template.
`templates/c/Makefile`	Adds `env`, `env-script`, `help` targets for C SDK template.
`templates/cxx/Makefile`	Adds `env`, `env-script`, `help` targets for C++ SDK template.
`templates/README.md`	Documents new template Makefile targets and usage.
`tests/lint/test-makefile-env-target.sh`	Validates `make env` prints eval-able exports.
`tests/lint/test-makefile-help-env-targets.sh`	Validates `make help` documents `make env` usage.
`tests/sdk/test-template-env-targets.sh`	Validates templates include the new Makefile targets and exports.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-02-08T12:09:36Z

tests/sdk/test-template-env-targets.sh

+  if ! grep -q 'env' "$MAKEFILE" | head -1; then
+    echo "FAIL: $lang/Makefile .PHONY missing env"


The .PHONY check is broken: grep -q produces no output, so piping it to head -1 makes the pipeline succeed regardless of whether the match exists. This can cause the test to miss missing .PHONY entries. Suggest checking the .PHONY: line explicitly (e.g., grep for ^\.PHONY: containing env env-script help) without piping.

Suggested change

if ! grep -q 'env' "$MAKEFILE" | head -1; then

echo "FAIL: $lang/Makefile .PHONY missing env"

if ! grep -q '^\.PHONY:.*\benv\b.*\benv-script\b.*\bhelp\b' "$MAKEFILE"; then

echo "FAIL: $lang/Makefile .PHONY missing env, env-script, or help"

Copilot · 2026-02-08T12:09:37Z

tests/lint/test-makefile-env-target.sh

+# Verify the output is valid shell (can be eval'd without error)
+eval "$ENV_OUTPUT" 2>/dev/null
+if [ $? -ne 0 ]; then
+  test_fail "make env output is not valid shell syntax"
+fi
+
+# Verify AGENTIZE_HOME was actually set after eval
+if [ -z "$AGENTIZE_HOME" ]; then
+  test_fail "AGENTIZE_HOME not set after eval \$(make env)"
+fi


This test asserts AGENTIZE_HOME is set after eval, but AGENTIZE_HOME is already exported by tests/common.sh, so this check can pass even if make env failed to set it. Consider unset AGENTIZE_HOME (and possibly PYTHONPATH) before eval, then assert the variables are set to the expected values afterwards.

Copilot · 2026-02-08T12:09:37Z

templates/python/Makefile

+env:
+	@echo 'export PROJECT_ROOT="$(CURDIR)"'
+	@echo 'export PATH="$(CURDIR)/build/bin:$$PATH"'
+	@echo 'export PYTHONPATH="$(CURDIR)/src:$$PYTHONPATH"'
+
+env-script:
+	@echo "Generating setup.sh..."
+	@echo '#!/bin/bash' > setup.sh
+	@echo '# Generated by make env-script' >> setup.sh
+	@echo 'export PROJECT_ROOT="$(CURDIR)"' >> setup.sh
+	@echo 'export PATH="$$PROJECT_ROOT/build/bin:$$PATH"' >> setup.sh
+	@echo 'export PYTHONPATH="$$PROJECT_ROOT/src:$$PYTHONPATH"' >> setup.sh
+	@chmod +x setup.sh


PYTHONPATH is being set to $(CURDIR)/src, but this template doesn't have a src/ directory (it has the package directory at the repo root, e.g. project_name/). As a result, eval $(make env) / make env-script won't actually put the SDK on PYTHONPATH. Update the export to match the template layout (likely $(CURDIR) or a configurable source path).

Address code review findings: - Makefile: fix shell escaping in help target so `make help` does not execute `make env` as a side effect (use `\$$` instead of `$$`). - tests: add executable permission to all 3 new test files. - tests/sdk/test-template-env-targets.sh: fix dead test check where `grep -q | head -1` always succeeds; use proper .PHONY line check. Test status: 60/60 shell tests passed. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

…16) scripts/mega-planner.py: new standalone 7-stage multi-agent debate pipeline script with full 4-mode CLI (default, --from-issue, --refine-issue, --resolve-issue). Uses only agentize.workflow.api (Session DSL, run_acw, prompt/path/gh utils). scripts/mega-planner.md: documentation for the mega-planner script interfaces. scripts/prompts/: co-located agent prompt files (7 files copied from .claude-plugin/). scripts/prompts/README.md: directory documentation explaining prompt file sources. scripts/README.md: added mega-planner section to scripts directory documentation. python/tests/test_mega_planner.py: 12 tests covering pipeline stages, execution order, prompt rendering, resolve mode, skip_consensus, and feature name extraction. python/agentize/workflow/planner.py: deleted deprecated backward-compat shim (shadowed by planner/ package directory, marked TODO:Delete). python/agentize/workflow/planner.md: deleted documentation for deprecated shim. All 34 tests pass (12 new mega-planner + 22 existing planner). Implements the plan from issue #16.

scripts/mega-planner.py: fix debate report header to list only 5 perspectives (items 6-7 are appended dynamically by caller); pass full feature_desc to consensus prompt instead of truncated feature_name; narrow except clause to specific exception types; replace Optional[str] with str | None for consistency with PEP 604 style.

Tonny-Gu added the agentize:pr PR created by agentize workflow label Feb 8, 2026

Copilot AI review requested due to automatic review settings February 8, 2026 12:04

Copilot started reviewing on behalf of Tonny-Gu February 8, 2026 12:04 View session

Copilot AI reviewed Feb 8, 2026

View reviewed changes

Tonny-Gu and others added 3 commits February 8, 2026 15:10

Tonny-Gu changed the title ~~[#16][feat] Add make env and env-script targets for environment setup~~ [feat][#16] Re-implement mega-planner as standalone Python script Feb 8, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[feat][#16] Re-implement mega-planner as standalone Python script#17

[feat][#16] Re-implement mega-planner as standalone Python script#17
Tonny-Gu wants to merge 4 commits intomainfrom
issue-16

Tonny-Gu commented Feb 8, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Feb 8, 2026

Uh oh!

Copilot AI Feb 8, 2026

Uh oh!

Copilot AI Feb 8, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

		if ! grep -q 'env' "$MAKEFILE" \| head -1; then
		echo "FAIL: $lang/Makefile .PHONY missing env"

Conversation

Tonny-Gu commented Feb 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

[feat][#16] Re-implement mega-planner as standalone Python script

Summary

Changes

Testing

Related Issue

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Tonny-Gu commented Feb 8, 2026 •

edited

Loading