epoch_forge/docs/roadmap.md

46 lines
2.1 KiB
Markdown

# Roadmap
The detailed backlog is [AGENTS-TASKS.md](../AGENTS-TASKS.md).
## Implemented foundation
- Local CLI repository with full-world epoch snapshots.
- Five entity collections with add/edit/show/diff/delete operations.
- Statistics, positions, duration/effects, selected relationships, and random
generation.
- Timeline advance, switch, log, tags, summaries, and explicit future deletion.
## Current release target
The trustworthy-foundation milestone is complete: documentation now reflects
the executable, user-triggerable panic paths found in the audit are guarded,
initialisation is safe, `show`, read-only `fsck`, and world diff are real,
individual files are persisted atomically, and CLI workflows are covered.
The dependable-daily-CLI milestone is complete: naming and aliases are
consistent, organisations are managed directly, help includes examples,
Nerd Font output has a portable build option, relationships are type-validated,
and every mutation has documented and tested recovery guidance.
The [Phase 4 format contract](staging-design.md) is approved. Versioned format
types, legacy fixtures, validation, dry-run migration, durable transient-state
creation, and byte-for-byte manifest backup are implemented. Format 2 remains
opt-in and ordinary commands refuse it until transaction locking and staging
land. Organisation-agnostic custom fields and types follow reversible history,
with unknown-field retention designed before that format change.
## Proposed stretch goals
1. Shell completions and generated man pages.
2. Stable JSON/JSONL output and query/filter expressions.
3. A keyboard-driven TUI over the same domain API (no GUI).
4. Named timeline branches with merge/conflict handling.
5. Setting templates for empires, companies, guilds, and nations.
6. Validation/simulation hooks before committing an epoch.
7. Optional signed snapshots and append-only audit records.
8. Deterministic seeded generation/simulation.
9. A scripting/plugin boundary after persistence is stable.
Completions and structured output are the best first stretch investments. A TUI
and branching should wait until staging, revert, and locking are reliable.