cli for Discord with sqlite backend
Pre-1.0 by version number (v0.7.2) but advancing rapidly. Created 2026-03-07 — only ~65 days old at time of appraisal — yet already at a seventh minor release with a tag cut the day before last commit. The feature surface (bot sync, wiretap, Git snapshot publishing, TUI, FTS5, AI-written field notes) is wide for a two-month-old project, raising a mild concern about depth versus breadth, but the 0 open issues and tight release cadence suggest it's tracking quality actively. Beta-stable is the right characterization; not yet production-hardened.
README is thorough and well-structured: requirements, bot permission checklist, token resolution order with OS-specific keyring commands, XDG path table, upgrade/migration notes, quick-start, and per-command reference stubs. References a docs/commands/tui.md file, indicating docs extend beyond the README. Missing: dependency manifest, CI badge, test coverage indication, and an explicit changelog (only inferred from release tags). Homebrew tap install path is a plus for discoverability. Gets new users to a working discrawl doctor run with minimal guesswork.
Go is a well-matched language for a CLI/SQLite pipeline: single-binary distribution, strong stdlib, solid database/sql ecosystem. Technology choices are sound — SQLite with FTS5 for full-text search, OS keyring integration, XDG Base Directory compliance, and a Gateway tail for live updates show architectural thoughtfulness. However, the dependency manifest is unavailable, so dependency hygiene, test presence, and CI status cannot be verified directly. No test files or CI workflow are referenced in the README. The wide feature set in a short timeframe is a code-quality risk until test coverage can be confirmed.
Last commit 2026-05-11 — one day before appraisal. Release v0.7.2 cut the same day. Zero open issues. The project shows daily or near-daily activity in its most recent window, and the release cadence (seven minor bumps in ~65 days) indicates an author who ships iteratively and triages promptly.
732 stars and 72 forks in roughly two months is a healthy early trajectory — approximately 11 stars/day average, which places it well above noise for a niche CLI tool. Fork count at ~10% of stars suggests genuine downstream use rather than passive interest. No downstream dependent data is available from the provided manifest, but the Homebrew tap reference implies the author expects distribution beyond source-only users.
Overall: 3.7/5
Category: Discord History Archiving
Known alternatives in vault: None. No existing vault entry targets Discord archiving, chat log mirroring, or guild data extraction.
Differentiation: discrawl occupies a specific and underserved niche: local-first, bot-token-based Discord guild archiving into a queryable SQLite store with FTS5 search, plus a credential-free wiretap path for local Desktop cache DMs. The Git-backed snapshot publish/subscribe pattern is a notable architectural differentiator — teams can distribute a read-only org archive without granting bot credentials to every reader. No vault alternative offers any of this. The TUI browser and --json machine interfaces make it automation-friendly beyond just a dump tool.
Gap or crowd: Clear gap. This is the sole entry in its functional category. Vault coverage is thin across most categories (10 of 11 categories have exactly one repo), and Discord archiving does not appear in any existing category. Adds net-new capability with zero redundancy.
Score: 3/5
Harvestable: SQLite schema design for multi-guild message/member/thread storage; FTS5 indexing pattern for append-heavy message logs; Git snapshot publish/subscribe pattern for credential-free archive distribution; --json flag convention for machine-readable status (metadata --json, status --json, doctor --json); OS keyring resolution order (env → keyring) as a portable secret-management pattern.
Integration path: discrawl can act as a structured data source for a PAI knowledge vault — Discord server history becomes a locally queryable corpus via direct SQL or the exposed read-only SQL interface. The search command could be wrapped as a PAI tool/skill; the SQLite file itself could be ingested by an embedding pipeline for semantic retrieval. The tail Gateway mode enables event-driven hooks (new messages → PAI ingest trigger). Git snapshot mode allows a PAI reader node to stay updated without Discord credentials.
Overlap with existing: No functional overlap with any vaulted repo. garrytan--gbrain and NorthwoodsSentinel--loam occupy personal AI memory but operate on different data substrates (general notes/documents, not chat logs). No collision.
Adoption cost: Moderate. Requires a Discord bot provisioned with correct intents and guild access, plus initial sync --full which may be time-consuming for large guilds. Integration with a PAI pipeline then requires writing an ingest adapter against the SQLite schema (no official PAI SDK hook exists). Wiretap path is trivial to enable on machines with Discord Desktop. Git-only reader mode has the lowest adoption cost of any path.
discrawl is a well-scoped, actively maintained CLI that fills a genuine gap: converting ephemeral Discord guild history into a durable, searchable, locally owned SQLite corpus. The architecture is sensible (Go + SQLite + FTS5 + Git snapshot distribution) and the documentation is above average for a two-month-old project. The primary PAI relevance is as a chat-corpus data source rather than a reasoning or retrieval component itself — its value scales with how central Discord is to the user's actual communication and knowledge workflows. For teams or individuals who live in Discord, this is an obvious capture layer; for users whose primary knowledge channels are documents, notes, or email, it adds marginal utility. The wiretap DM import path and zero-credential Git reader mode are architecturally interesting patterns worth studying even in non-Discord contexts. Recommend adding to vault as a thin-but-unique category entry; revisit at v1.0 for maturity re-score.