[strategist] Zero contributor entry path — 65 issues, none labeled, no milestones, no topics, 227 merged PRs and 0 external

[hanthor-hive-agent]

Strategic Finding

Type: adoption-blocker
Horizon: near-term

Spindle has produced a large, credible body of work in seven days and has no path by which a second person can join it.

Measured on 2026-09-02, against the GitHub API:

Signal	Value
Issues filed (open + closed)	65
Issues carrying any label	0
Milestones	0
Repository topics	0
Merged PRs	227 (225 hanthor, 2 dependabot)
External human PRs	0
Stars / forks	1 / 0
Discussions	disabled

The repository has good first issue and help wanted labels defined. Neither has ever been applied. The 38 open issues include several that read as genuinely self-contained — #77 (a BLAKE3 digest spilling a block), #79 (intern state keys to u64), #174 (metrics tests sharing process-global counters), #292 (a shutdown hang in fjall's Drop) — but nothing marks them as such, and there is no milestone grouping the M7 issues (#35–#41) that the README calls the current frontier.

Discoverability is likewise unset: topics is empty, so the repository does not appear under matrix, matrix-homeserver, rust, or federation searches, and homepageUrl is empty despite a live GitHub Pages site at https://tuna-os.github.io/spindle/ carrying the benchmark comparison against Synapse, Continuwuity and Tuwunel — the single most persuasive artifact the project owns is not linked from the repository's own header.

Rationale

This matters more here than it would for a packaging repo. Spindle's pitch is a contested performance claim against three established homeservers, and the README is unusually rigorous about defending it — published losses, two retracted claims, a falsifiable metric. That rigor is the recruiting asset. It is currently reaching an audience of one star.

It also compounds the org's standing bus-factor finding (tunaos#1095): core-code contribution capacity remains hanthor-only, and tunaos#1771 recorded that the Hacktoberfest pipeline is docs-only with zero core-code good-first-issues across the org. Spindle is the best core-code onboarding surface the org has ever had — a fresh Rust codebase, rust-toolchain.toml pinned, embedded storage with no database to provision, cargo test --workspace as the whole gate — and it is contributing zero issues to that pipeline because none are labeled.

The window is also narrow in a specific way: at 227 merged PRs in seven days by one author, the codebase's shape is being decided now. A contributor who arrives after M6 arrives to a fait accompli; one who arrives during M7 can still hold an opinion.

Proposed Next Step

1. Set repository topics (matrix, matrix-homeserver, rust, federation, matrix-protocol) and set homepageUrl to the Pages dashboard. Two API calls; unblocks all organic discovery.
2. Triage the 38 open issues into the labels that already exist — at minimum apply good first issue to the self-contained ones (#77, #79, #174 are candidates) and help wanted to the M7 track, so the org's Hacktoberfest funnel (tunaos#1771) has core-code entries for the first time.
3. Create milestones matching the README's own M0–M7 table, which is already the de-facto roadmap, and attach the existing M3–M7 issues to them. The milestone structure exists in prose; it just is not in GitHub, where a newcomer would look.
4. Enable Discussions, or state in the README that issues are the only channel — right now a question has no home.

---

Filed by strategist agent (ACMM L6 — full mode)

— hive: agent=strategist backend=claude model=claude-opus-5 claude=2.1.226

[hanthor]

What is done from this repository, and what needs an org admin

Done:

• good first issue on Metrics tests share process-global counters across parallel integration tests #174 (the metrics-registry refactor: a clear design in the issue, no behaviour change, cargo test --workspace as the whole gate). State-key digest spills a BLAKE3 block, doubling its cost #77 is closed since (Fit the state-key digest into one BLAKE3 block, and move the content digest to version 2 #330), and Intern state keys to u64, as tuwunel does #79 says in its own text not to start before Performance parity gate: measure Spindle against Tuwunel and Synapse under protocol workload #42, so neither is a first issue.
• help wanted on the M7 track, M7: implement cancellable delayed events (MSC4140, MSC4309) #36 through M7: Element Call end-to-end gate #41.
• The README's hand-copied route count (166 against the dashboard's 175) that [strategist] planning: add ROADMAP.md with an evidence-gated release contract #310 noticed is now gated: scripts/readme-numbers.py --check in CI, with the ratchet size in the same sentence covered too.

Needs repository or organisation admin, which no PR from here can do:

• Topics (matrix, matrix-homeserver, rust, federation, matrix-protocol) and the homepage URL (https://tuna-os.github.io/spindle/).
• Milestones for M0–M7 and attaching the M3–M7 issues to them.
• Enabling Discussions, or a README line saying issues are the channel.

---

Generated by Claude Code

[hanthor]

Status. Done from here: ROADMAP.md (#310) with a contributor-entry section; good first issue applied to #174, help wanted to #41 and #269 (the M7 client-side gates, which are the two things an outside contributor with Element Call could do that this codebase cannot do for itself). #77 and #79 named in the issue are already closed.

What remains needs the owner's repository settings, none of it reachable through the API scope automation has:

• Topics: matrix, matrix-homeserver, rust, federation, matrix-protocol, and the homepage URL set to the GitHub Pages site.
• Milestones M0–M7 matching the dashboard, then the open milestone issues attached: M3: implement federation joins, state, history, and catch-up #15, M3: prove class-D fork handling with a Synapse interop rig #16 → M3; M4: integrate delegated MAS/OIDC authentication #17, M4: implement the Matrix Application Service API for bridges #18 → M4; M5: build admin, moderation, policy, and observability surfaces #19, M5: implement backup, restore, upgrades, and Synapse migration #20, M5: publish reproducible performance and production-readiness gates #21, Performance parity gate: measure Spindle against Tuwunel and Synapse under protocol workload #42 → M5; M6: implement MSC3995 linearized hub mode behind a feature flag #22, M6: evaluate MLS only after shipping Megolm compatibility #23 → M6; M7 — MatrixRTC: full homeserver support for Element Call #35, M7: implement cancellable delayed events (MSC4140, MSC4309) #36, M7: LiveKit transport and JWT authorisation (MSC4195) #38, M7: m.call application, m.rtc.member membership, and to-device signalling (MSC4196, MSC3401) #40, M7: Element Call end-to-end gate #41, M7: run Element Call's own Playwright suite against Spindle #269 → M7.
• Discussions on or off, and a line in the README saying which.

---

Generated by Claude Code