Issue-128: Standardize python3 invocation across skill instructions and scripts

[jodavis-claude]

Work item

Issue-128: Skills and scripts that shell out to Python assumed a bare python command exists, but several real environments (including this devcontainer) only provide python3. This task standardizes all agent-facing invocation examples and generated command strings on python3 / sys.executable.

Changes

• Replaced bare python with python3 in all Bash invocation examples across plugins/dev-team/skills/*/SKILL.md (concurrent-orchestrate, dev-spec-task-breakdown, ensure-working-branch, get-project-configuration, implement-tdd, monitor-pr, update-project-configuration, use-context-file, workflow-orchestrate, workflow-script), including the top-level orchestrator launch command in workflow-orchestrate/SKILL.md.
• Updated scripts/dev_team_update.py's usage docstring for consistency.
• Changed dev_team.py's two generated shell-command strings (_resolve_validation_script() and BuildValidationStep.get_actions()) to use sys.executable instead of hardcoding python, matching the repo's existing pattern for one Python process spawning another.
• Added unit test coverage asserting the emitted commands start with sys.executable rather than a hardcoded python prefix.

Design decisions

• Standardized fully on python3/sys.executable rather than adding a python-then-python3 detection/fallback, per the issue author's own comment narrowing scope to "assume python3 is the correct runtime," and consistent with every .py script in the repo already declaring #!/usr/bin/env python3.
• Did not add a "check python3 exists up front, offer to install" preflight step — the issue comment hedged this as a "maybe," and it is a materially larger, riskier scope (sudo/package-manager access, cross-platform logic) than the rest of this fix.
• Left _spec_AgentOrchestration.md's illustrative python dev_team.py ... examples unchanged, since it is a historical/superseded design spec rather than the authoritative runtime instructions agents follow.

Closes#128

Testing completed

• Existing unit tests in test_dev_team.py for _resolve_validation_script and BuildValidationStep.get_actions continue to pass unchanged.
• Added new unit tests asserting generated commands use sys.executable instead of a bare python prefix.

[github-actions]

build-and-test: Python test results

Status: ✅ Passed

Test log

[jodavis-claude]

Review summary

Reviewed the diff against Issue-128's exit criteria: standardize all Python invocations on python3/sys.executable.

What's good:

• All 11 SKILL.md files listed in the task brief were updated; no bare python invocation remains anywhere under plugins/dev-team/ (verified with a repo-wide grep).
• dev_team.py's two generated command strings (_resolve_validation_script, BuildValidationStep.get_actions) now use sys.executable, matching the repo's existing pattern for one Python process spawning another (sys was already imported).
• New unit tests (test_validation_list_resolves_to_sys_executable_command, TestBuildValidationStepGetActions::test_run_script_command_uses_sys_executable) assert on the sys.executable prefix. Ran the full suite locally: 143 passed.
• scripts/dev_team_update.py's usage docstring and workflow-script/SKILL.md's example were updated for consistency.
• CI (build-and-test, devcontainer-validate, gate) all pass.

Priority 1 — Correctness / missed instance of the exact bug (blocking):

hooks/hooks.json (not touched by this PR) still invokes the SessionStart update script with a bare python:

"command": "python \"${CLAUDE_PLUGIN_ROOT}/scripts/dev_team_update.py\" --data-dir \"${CLAUDE_PLUGIN_DATA}\" --threshold-hours 4"

This is real, automatically-executed configuration (not agent-facing prose) — it runs on every Claude Code SessionStart for this plugin, before any skill or agent logic runs. In an environment with only python3 on PATH (confirmed reproducible in this exact devcontainer per the task brief: which python → not found, exit 127), this hook will fail with the very "command not found: python" error the issue is about — arguably the single highest-impact remaining instance, since it fires unconditionally rather than only when a particular skill happens to be invoked.

Since this runs standalone (no controlling Python process to source sys.executable from), the fix should mirror the rest of this sweep and hardcode python3 here too:

"command": "python3 \"${CLAUDE_PLUGIN_ROOT}/scripts/dev_team_update.py\" --data-dir \"${CLAUDE_PLUGIN_DATA}\" --threshold-hours 4"

Recommend adding this file to the PR before merging — it falls squarely within the issue's own scope ("Skills and scripts that shell out to Python...") even though it wasn't enumerated in the task brief's file list.

No other Priority 1-4 issues found. Documentation: no _doc_*.md mentions a python-invocation convention, so nothing there needed updating. CONTRIBUTING.md's "Supported Platforms" section was left without an explicit python3-pinning note (task brief's "Known ambiguity #5") — reasonable to leave as a non-blocking follow-up since the issue didn't require it.

Style (non-blocking): none noted.