Squad product: Linked chat summaries -- investigation and StorageProvider dependency

[diberry]

Scope: This is a Squad product investigation issue. It connects Brady's "Linked chat summaries" feature exploration to the StorageProvider abstraction in PR bradygaster#640.

Background: Brady's "Linked chat summaries" email thread

Brady initiated an email thread exploring linked chat summaries as a new Squad feature. Key points from the discussion (Brady, Paul Hsu, and team):

• Treat linked chat summaries as an experimental, gated feature
• Develop and test in a private fork/branch first
• Use environment variables to gate the feature and tune thresholds
• Run established benchmarks (LOCOMO, LongMemEval, Tau Bench) to validate
• Collect opt-in telemetry from real users (full conversation histories, token usage, memory retrieval)
• Use simulation to compare outcomes when full dialog history is available
• Coordinate release messaging carefully
• Short testing window (a few days) before convergence

Connection to bradygaster#640 (StorageProvider)

PR bradygaster#640 introduces the StorageProvider interface with FSStorageProvider, InMemoryStorageProvider, and SQLiteStorageProvider. This is directly relevant to linked chat summaries because:

1. Chat history persistence -- linked chat summaries need to store and retrieve conversation histories. The StorageProvider abstraction gives this feature a clean persistence layer without coupling to the filesystem.

2. InMemoryStorageProvider for testing -- Paul's benchmarking plan (LOCOMO, LongMemEval, Tau Bench) needs fast, isolated test runs. InMemory provider enables this without disk I/O overhead.

3. SQLiteStorageProvider for telemetry -- opt-in telemetry collection (token usage, memory retrieval metrics) naturally fits a structured storage backend like SQLite rather than flat files.

4. Environment variable gating -- the StorageProvider pattern already supports config-driven selection. The linked chat summaries feature could use a similar pattern: SQUAD_CHAT_SUMMARIES=enabled selects the right storage backend.

5. Private fork development -- Brady wants to develop this in a private fork. The StorageProvider PR (feat(sdk): StorageProvider abstraction — complete migration + example providers bradygaster/squad#640) should land on upstream dev first so the private fork can build on it.

Work to be done

Pre-requisites (must land first)

• Merge feat(sdk): StorageProvider abstraction — complete migration + example providers bradygaster/squad#640 (StorageProvider) to upstream dev -- provides the persistence abstraction
• Strip bleed files from feat(sdk): StorageProvider abstraction — complete migration + example providers bradygaster/squad#640 (5 files identified by Flight's review)

Investigation tasks

• Determine where chat summaries would be stored (which StorageProvider implementation)
• Design the linked summary data model (what gets stored, how summaries link across sessions)
• Define the environment variable gating pattern (on/off switch, threshold tuning)
• Identify which benchmarks (LOCOMO, LongMemEval, Tau Bench) are feasible to run locally
• Design the opt-in telemetry schema (token usage, memory retrieval, conversation length)

Implementation tasks (after investigation)

• Implement chat summary storage using StorageProvider interface
• Add environment variable gates
• Build benchmark harness for comparison testing
• Add telemetry collection (opt-in, using StorageProvider)
• Create private fork branch for development

Process

• Align with Brady and Tamir on scope before cutting private fork
• Set up Teams thread for async coordination (Brady requested this)
• Define short testing window for experimentation

Related

• feat(sdk): StorageProvider abstraction — complete migration + example providers bradygaster/squad#640 -- StorageProvider abstraction (PR on bradygaster/squad, draft)
• Brady's email thread: "Linked chat summaries" (Brady, Paul Hsu)
• Squad product: Default workflows burn too many Actions minutes for multi-repo customers #98 -- Config-driven CI (environment variable gating pattern is shared)

[github-actions]

📋 Assigned to Flight (Lead)

Issue: #101 — Squad product: Linked chat summaries -- investigation and StorageProvider dependency

Flight will pick this up in the next Copilot session.

For Copilot coding agent: If enabled, this issue will be worked automatically.
Otherwise, start a Copilot session and say:
Flight, work on issue #101

[github-actions]

🏗️ Squad Triage — Flight (Lead)

Issue: #101 — Squad product: Linked chat summaries -- investigation and StorageProvider dependency
Assigned to: FIDO (Quality Owner)
Reason: Issue relates to testing/quality work

@copilot evaluation: 🔴 Not suitable — issue involves work outside the coding agent's capability profile.

Team roster:

• Flight (Lead) → label: squad:flight
• Procedures (Prompt Engineer) → label: squad:procedures
• EECOM (Core Dev) → label: squad:eecom
• FIDO (Quality Owner) → label: squad:fido
• PAO (DevRel) → label: squad:pao
• CAPCOM (SDK Expert) → label: squad:capcom
• CONTROL (TypeScript Engineer) → label: squad:control
• Surgeon (Release Manager) → label: squad:surgeon
• Booster (CI/CD Engineer) → label: squad:booster
• GNC (Node.js Runtime) → label: squad:gnc
• Network (Distribution) → label: squad:network
• RETRO (Security) → label: squad:retro
• INCO (CLI UX & Visual Design) → label: squad:inco
• GUIDO (VS Code Extension) → label: squad:guido
• Telemetry (Aspire & Observability) → label: squad:telemetry
• VOX (REPL & Interactive Shell) → label: squad:vox
• DSKY (TUI Engineer) → label: squad:dsky
• Sims (E2E Test Engineer) → label: squad:sims
• Handbook (SDK Usability) → label: squad:handbook
• Ralph (Work Monitor) → label: squad:ralph
• @copilot (Coding Agent) → label: squad:copilot

To reassign, remove the current squad:* label and add the correct one.