v0.7.0 — /p2e-add-story --thick + brainstorming escalation (B-05-L19)

v0.7.0 — 2026-04-18

Adds opt-in --thick mode to /p2e-add-story and wires a bounded brainstorming escalation into both /p2e-add-story --thick and /p2e-update-story thicken. Additive; the default /p2e-add-story invocation and every existing /p2e-update-story path are unchanged.

Added

• /p2e-add-story --thick — new opt-in flag that populates ALL thick-spec fields at add time (the same six fields /p2e-update-story thicken populates: filesHint, constraints, nonGoals, contextDocs, effortHint, verificationCmd), runs the sizing inference heuristic against the staged projection per workflows/p2e-sizing-rubric.md, and renders the annotated preview with provenance labels on every field. Thin mode (the default) is unchanged.
• Sizing inference at add time (thick mode only) — the drafter runs the rubric's inference inputs (title + capabilities + AC count + tags + files_hint length) and annotates the proposed tier derived-from-source: <evidence> instead of the thin-mode defaulted-M. The user may still override the inferred tier in the confirm step's Adjust sizing action.
• Brainstorming escalation in both /p2e-add-story --thick and /p2e-update-story thicken — when the source signal is insufficient to credibly fill ≥ 2 thick-spec fields, the wrapper invokes the host brainstorming primitive (superpowers:brainstorming on Claude; Codex's native equivalent) to batch 2–4 concrete questions in a single turn. Answers fold back into the staged draft before preview re-render, annotated derived-from-brainstorming. Single round per flow; never bypasses the preview/confirm gate. Empty cells are still preferred over filler when answers leave gaps.
• --thick-mode confirm step extensions in workflows/p2e-add-story.md — the confirm step now supports adjusting any of the six thick-spec fields inline, with the override annotated steered-by-user in the re-rendered preview.
• New Draft a thick P2E story from this feature idea entry in the Codex defaultPrompt list (.codex-plugin/plugin.json).
• Validator coverage in scripts/validate-plugin.py (validate_thick_mode_contract) asserting the new --thick, ## Modes, ## Brainstorming escalation, and derived-from-brainstorming phrases exist on the expected surfaces (both workflows, both commands, both skills).

Changed

• workflows/p2e-add-story.md — adds ## Modes, augments the ## Workflow steps to branch on thick vs thin, extends ## Required preview contents and ## Required confirm step for the six thick-spec fields and the derived-from-brainstorming provenance label, rewrites ## Sizing rules to cover both modes, and adds ## Brainstorming escalation with explicit escalation-trigger + fold-back rules.
• workflows/p2e-update-story.md — adds ## Brainstorming escalation with the same shared contract, and extends the sizing-row provenance set with derived-from-brainstorming.
• commands/p2e-add-story.md — argument hint adds [--thick]; body describes thin vs thick mode + the brainstorming escalation.
• commands/p2e-update-story.md — body adds a Brainstorming escalation paragraph pointing at the workflow contract.
• skills/p2e-add-story/SKILL.md + skills/p2e-update-story/SKILL.md — hard-rule blocks cover the thick-mode inference path and the bounded brainstorming escalation.
• .claude-plugin/marketplace.json + .codex-plugin/plugin.json versions bumped to 0.7.0.

Notes

• No breaking changes. The default /p2e-add-story <description> invocation stays thin; --thick is purely opt-in.
• Brainstorming escalation is bounded to one round per flow and only fires when ≥ 2 thick-spec fields would otherwise land empty. It never bypasses the preview/confirm gate. The Claude wrapper resolves the reference against superpowers:brainstorming; the Codex wrapper resolves it against its native brainstorming primitive (the same pattern already used by workflows/p2e-bootstrap.md --mode=onboarding).
• python3 scripts/validate-plugin.py passes.