mirror of
https://github.com/rennf93/roboco.git
synced 2026-08-03 07:23:24 +02:00
5b27a443e9
Thread the deployer's product name through the X reply + feature-spotlight prompts (B6 leftover; release/video paths shipped in #570); make the docs-site repo/URL config (ROBOCO_DOCS_SITE_*, defaults unchanged) instead of a roboco-website hardcode (B8); de-assert our repo from the Main PM prompt (B10); derive PR labels from the real target branch instead of literal to-master/to-slave; drop the stale headcount from base.md; and make the bash-guard's Makefile check require an actual quality/gate/lint/ test target before denying raw package-manager commands (no more false-remediation loop on Go/Rust Makefiles). Co-authored-by: Renn F <rennf93@users.noreply.github.com>
130 lines
5.9 KiB
Python
130 lines
5.9 KiB
Python
"""Cluster C4 — agent briefing/onboarding fixes (findings #3, #5, #9).
|
|
|
|
These assert on the COMPOSED system prompt (``compose_prompt``), which is
|
|
the text actually mounted into agent containers at ``/app/system-prompt.md``
|
|
via ``--system-prompt-file``. The three behaviors fixed here:
|
|
|
|
- #3: Main PM must read the upstream PO/HoM handoff BEFORE its own research,
|
|
so it does not duplicate the Board's analysis.
|
|
- #5: Main PM's Products-vs-Projects mental model — a Product fans out to
|
|
per-cell Projects which may be the SAME repo (monorepo subtrees) or
|
|
DIFFERENT repos (multi-repo); the Main PM coordinates across them and
|
|
must not assume one repo.
|
|
- #9: Developers know their exact workspace path convention, stay inside
|
|
their own cell workspace, and obtain secrets via the task description —
|
|
never ``env``/``printenv`` (bash-guard denies it).
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
from roboco.agents.factories._base import compose_prompt
|
|
from roboco.models import AgentRole, Team
|
|
|
|
|
|
def _composed_prompt_for(role: AgentRole, team: Team | None = None) -> str:
|
|
return compose_prompt(role, team, agent_slug="test-agent")
|
|
|
|
|
|
# --------------------------------------------------------------------------
|
|
# #3 — Main PM consumes the upstream handoff before researching/planning
|
|
# --------------------------------------------------------------------------
|
|
|
|
|
|
def test_main_pm_prompt_requires_reading_upstream_handoff_first() -> None:
|
|
"""Main PM is told to read the PO/HoM handoff BEFORE its own research."""
|
|
prompt = _composed_prompt_for(AgentRole.MAIN_PM)
|
|
assert "Read the upstream handoff BEFORE you research or plan" in prompt
|
|
# Names both upstream sources explicitly.
|
|
assert "Product Owner" in prompt
|
|
assert "Head of Marketing" in prompt
|
|
|
|
|
|
def test_main_pm_prompt_forbids_re_researching_already_handed_work() -> None:
|
|
"""The duplicated-research failure mode is called out as an anti-pattern."""
|
|
prompt = _composed_prompt_for(AgentRole.MAIN_PM)
|
|
assert "Re-researching the codebase from scratch" in prompt
|
|
# The handoff lives in the task journal as decision/reflect entries.
|
|
assert "decision" in prompt and "reflect" in prompt
|
|
|
|
|
|
# --------------------------------------------------------------------------
|
|
# #5 — Products-vs-Projects mental model (mono- vs multi-repo)
|
|
# --------------------------------------------------------------------------
|
|
|
|
|
|
def test_main_pm_prompt_explains_product_fans_out_to_per_cell_projects() -> None:
|
|
"""Main PM prompt distinguishes Product (strategic) from per-cell Projects."""
|
|
prompt = _composed_prompt_for(AgentRole.MAIN_PM)
|
|
assert "Products vs Projects" in prompt
|
|
assert "fans out to one Project per cell" in prompt
|
|
|
|
|
|
def test_main_pm_prompt_covers_both_monorepo_and_multirepo() -> None:
|
|
"""Both fan-out shapes are described; neither is assumed by default."""
|
|
prompt = _composed_prompt_for(AgentRole.MAIN_PM)
|
|
assert "Monorepo" in prompt
|
|
assert "Multi-repo" in prompt
|
|
# Must not let the agent call a monorepo subtree "a separate repo".
|
|
assert "not" in prompt and "a separate repo" in prompt
|
|
|
|
|
|
def test_main_pm_prompt_frames_monorepo_as_per_deployment_fact() -> None:
|
|
"""The monorepo case is illustrated generically, not asserted as a fact
|
|
about a specific repo — the Main PM is told to confirm the shape from
|
|
each cell's ``project_slug``, never assume one deployment's config
|
|
(this repo's own `rennf93/roboco`) applies to every Product."""
|
|
prompt = _composed_prompt_for(AgentRole.MAIN_PM)
|
|
assert "per-deployment fact you confirm from each cell's" in prompt
|
|
assert "github.com/rennf93/roboco" not in prompt
|
|
|
|
|
|
# --------------------------------------------------------------------------
|
|
# #9 — Developer workspace path convention + secret handling
|
|
# --------------------------------------------------------------------------
|
|
|
|
|
|
def test_developer_prompt_states_exact_workspace_path_convention() -> None:
|
|
"""Developer prompt gives the exact /data/workspaces/<slug>/<team>/<slug> path."""
|
|
prompt = _composed_prompt_for(AgentRole.DEVELOPER, Team.BACKEND)
|
|
assert "/data/workspaces/<project-slug>/<team>/<agent-slug>/" in prompt
|
|
# Steered to stay inside its own cell workspace.
|
|
assert "Stay inside your own cell workspace" in prompt
|
|
|
|
|
|
def test_developer_prompt_forbids_probing_filesystem_for_workspace() -> None:
|
|
"""Developer is told not to probe / guess the workspace path."""
|
|
prompt = _composed_prompt_for(AgentRole.DEVELOPER, Team.BACKEND)
|
|
assert "Do NOT probe for it." in prompt
|
|
assert "ls /" in prompt and "find /" in prompt
|
|
|
|
|
|
def test_developer_prompt_gives_sanctioned_secret_path_not_env() -> None:
|
|
"""Secrets come via the task description; env/printenv is denied."""
|
|
prompt = _composed_prompt_for(AgentRole.DEVELOPER, Team.BACKEND)
|
|
# env/printenv called out as denied + bash-guard blocked.
|
|
assert "printenv" in prompt
|
|
assert "bash-guard" in prompt
|
|
# Sanctioned path: value provided in the task description.
|
|
assert "in the task description" in prompt
|
|
# Escalation route when the value is genuinely missing.
|
|
assert "i_am_blocked" in prompt
|
|
|
|
|
|
# --------------------------------------------------------------------------
|
|
# Main PM delegation description = brief (goal + constraints), not a solution
|
|
# --------------------------------------------------------------------------
|
|
|
|
|
|
def test_main_pm_prompt_delegation_description_is_a_brief() -> None:
|
|
"""The delegate description must be a goal+constraints brief, not a spec."""
|
|
prompt = _composed_prompt_for(AgentRole.MAIN_PM)
|
|
assert "The description is a **brief**, not a spec." in prompt
|
|
assert "Do NOT prescribe the cell's solution" in prompt
|
|
|
|
|
|
def test_main_pm_prompt_forbids_dictating_the_cells_design() -> None:
|
|
"""The design-task failure mode (PM dictating the UX layout) is called out."""
|
|
prompt = _composed_prompt_for(AgentRole.MAIN_PM)
|
|
assert "Design the UX and propose the layout" in prompt
|
|
assert "give them the problem, not your mockup" in prompt
|