v1.2.0 to git:20260905.c68a71a

33 added, 83 removed. Audit A to A.

---

name: arsenal

- description: Clarify a fuzzy project idea into a phased roadmap through a Socratic interview, capability-aware research, and adversarial review. Use for roadmap discovery, product direction, or deciding what to build before brief, proposal, or implementation work.

- compatibility: Python 3 is required only for HTML rendering. Network access and subagents are optional; the complete workflow has a solo fallback.

- metadata:

- version: "1.2.0"

+ description: Clarify a genuinely fuzzy idea into a concise phased roadmap. Optional bounded research, no automatic agents.

+ argument-hint: "<idea> [-a] [-r slug] [--no-agents] [--html]"

---

- <objective>

- Turn a vague idea into a clear phased roadmap the user believes in. Interview

- first, research existing solutions, draft the smallest useful sequence, review

- it adversarially, and render one durable roadmap artifact.

- </objective>

+ # Arsenal roadmap

- <parameters>

- | Flag | Description |

- |------|-------------|

- | `-a` / `--auto` | Infer interview answers from the idea |

- | `-r` / `--resume` | Reopen an existing roadmap for discussion |

- | `--no-agents` | Force the complete solo workflow |

- | `--install-omp-agents` | Explicitly install bundled OMP agents |

- | `--install-codex-agents` | Explicitly install bundled Codex agents |

- | `--uninstall-omp-agents` | Restore or remove installed OMP agents |

- | `--uninstall-codex-agents` | Restore or remove installed Codex agents |

+ Keep the existing roadmap schema and resume behavior. Read the requested

+ roadmap if resuming; do not restart a finished phase.

- The remaining input is the project idea or resume slug.

- </parameters>

+ Clarify the actual pain, intended user, smallest useful result, non-goals and

+ constraints only where missing. Ask a relevant plain-text question, not a

+ fixed interview quota. Stop interviewing when the direction is actionable.

- <state_variables>

- | Variable | Type | Set by |

- |----------|------|--------|

- | `{idea}` | string | step-00-init |

- | `{slug}` | string | step-00-init |

- | `{auto_mode}` | boolean | step-00-init |

- | `{resume_mode}` | boolean | step-00-init |

- | `{force_solo}` | boolean | step-00-init |

- | `{roadmap_dir}` | string | step-00-init |

- | `{capabilities}` | object | step-00-init |

- | `{answers}` | object | step-01-interview |

- | `{research}` | array | step-02-research |

- | `{execution_trace}` | array | steps 02 and 03 |

- | `{stepsCompleted}` | array | every step |

- </state_variables>

+ Run targeted research only where external evidence changes the roadmap.

+ Keep useful sources, what to borrow/avoid and uncertainty; no minimum count.

+ Use available native search capabilities. No web access means disclose the

+ limitation, not fabricate citations.

- <delimitation>

- - Arsenal decides what is worth building and in which order.

- - It does not write a product brief, technical proposal, or implementation.

- - The workflow hands off through semantic next actions. Harness adapters format

- any concrete command shown to the user.

- </delimitation>

+ Write docs/roadmap/<slug>/roadmap.md. Keep its six numbered sections from the

+ existing schema, but omit decorative content within them. Use as many phases

+ as the objective needs, no invented diagram or feature. Review in the lead.

- <entry_point>

- **FIRST ACTION:** Load `steps/step-00-init.md`.

- </entry_point>

+ Agents are optional: only after approval, and only genuinely independent

+ read-only units. Read references/agent-contracts.md and the matching adapter

+ only then. At most three; no fallback chain of agent launches. If a worker

+ fails, handle the missing part once in the lead. Do not open test sessions

+ to prove model routing during a roadmap task.

- <step_files>

- | Step | File | Purpose |

- |------|------|---------|

- | 00 | `steps/step-00-init.md` | Parse input and detect capabilities |

- | 01 | `steps/step-01-interview.md` | Find the real objective |

- | 02 | `steps/step-02-research.md` | Research through available capabilities |

- | 03 | `steps/step-03-plan.md` | Draft and adversarially review the roadmap |

- | 04 | `steps/step-04-render.md` | Render the HTML deliverable |

- | 05 | `steps/step-05-discuss.md` | Re-discuss, finalize, and hand off |

- </step_files>

+ Render scripts/render.py only with --html or an explicit request. Mark ready

+ when the user approves. Do not execute the next phase automatically.

- <references>

- - `references/capability-contract.md`: probe and fallback contract.

- - `references/agent-contracts.md`: bounded agent inputs and outputs.

- - `references/adapters/`: harness-specific discovery and routing.

- - `scripts/install_agents.py`: explicit OMP and Codex agent installation.

- - `scripts/render.py`: offline, cross-platform HTML renderer.

- </references>

+ Explicit install/uninstall flags remain supported: --install-omp-agents,

+ --install-codex-agents, --uninstall-omp-agents, --uninstall-codex-agents. Use

+ scripts/install_agents.py for the selected operation only, with backups.

+ They never run as a side effect of creating a roadmap. A restricted harness

+ may require the user to authorize this separate configuration action.

- <interaction>

- - The interview is interactive unless `--auto` is present.

- - Ask one plain-text question at a time.

- - Use the conversation language for chat and the roadmap. Keep versioned

- repository content, identifiers, commands, and frontmatter keys in English.

- - The lead owns interpretation, scope, synthesis, roadmap writing, and final

- discussion. Agents never question the user or write the roadmap.

- </interaction>

+ ## Execution policy

- <critical>

- - Resolve research, delegation, and rendering as semantic capabilities. Never

- require a particular tool name in the portable core.

- - Delegate only two or more independent, read-only units. Maximum fan-out is

- three. Run the same contracts sequentially in the lead when delegation is

- unavailable.

- - Preserve the roadmap schema and resume behavior. Optional execution trace

- metadata is additive.

- - Never modify global user configuration automatically. Agent installation is

- explicit and reversible.

- - Rendering the HTML is mandatory. Opening a browser is optional.

- - Write only `docs/roadmap/<slug>/roadmap.md` during a normal workflow.

- </critical>

+ Work solo. Ask before any subagent or reviewer, even in auto mode. Explain

+ the independent scope and expected benefit first. No hidden advisor, nested

+ delegation, model retuning, repeated successful checks, or progress spam.

+ Use existing context before asking questions. Stop when the requested result

+ is delivered. User stops and scope changes override pending steps.