ResearchSession Capture Research

20 — Comparison Matrix and Tier List

20 — Comparison Matrix and Tier List

Research conducted: 2026-07-03. Tools from 01/02 (SpecStory), 11 (OSS), 12 (platforms). Tier value = usefulness to a jackin session-capture effort — coverage of jackin's six-agent roster × mechanism robustness × output/search/share quality × maintenance health × license reusability. That is deliberately not "best product for an individual developer".

Capability matrix (the contenders)

Capture = how conversation content is acquired. Live = captures during the session vs post-hoc. Content = full conversation (incl. tool calls) vs metrics only.

ToolCaptureLiveContentOutputSearchShareTurn↔code linkLicenseHealth (2026-07-03)
SpecStory CLI v2native stores, fsnotify✅ (watch) + post-hocfull incl. thinking, tools, usagemarkdown in-repoFTS5, cross-projectcloud links (opt-in)prototype (provenance engine, hidden flag)Apache-2.0active, v2.0.0 4 days old
SpecStory IDE extCursor state.vscdb / Copilot chatSessions pollingfull (known gaps: Cmd-K, images)markdownvia cloudanon + cloud linksclosedactive; breakage history
Mantranative storesfull + per-turn git snapshotsdesktop timeline replay✅ FTSweb publishshipped (diff/restore)closed freewareactive (rel. 2026-07-01)
claude-replaynative stores (5 agents)post-hocfullself-contained HTML replayfile = artifactMITactive
coding_agent_session_searchnative stores (11+ providers)post-hocfull (indexed)TUI/CLI results✅ best-in-class breadthnonstandardvery active
claude-code-logClaude JSONLpost-hocfullHTML/MDproject browseOSSvery active
claude-code-transcriptsClaude JSONLpost-hocfullmulti-page HTMLgist / SSO sitesOSSslowing
aichat search (claude-code-tools)native stores (Claude, Codex…)post-hocfull (Tantivy index)CLI/TUIMITactive
claude-tracefetch-patch wire tapfull + system prompt + raw APIJSONL + HTMLfileOSSlow activity
Langfuse (hook route)Claude hooks → tracesfull conversations (this route only)web platformteam platformOSS core + SaaSactive
Happywraps CLI, E2E relaylive session statemobile/web viewerdevice sync, not linksMITvery active
vibe-kanbanowns launched sessionsper-attempt logsweb kanbanper-task diffsApache-2.0active
Amp threads (native)server-side by designfullweb threadsworkspace/unlisted URLsclosed SaaSactive
opencode /share (native)server publishfullweb pagepublic linksOSS agent, closed share svcactive
ccusage (for contrast)Claude JSONLpost-hocmetrics onlyCLI reportsleaderboardsMIT16.8k★, very active

Coverage vs jackin's roster

jackin runs Claude Code, Codex, Amp, Kimi, OpenCode, Grok (30). Coverage of that roster by the cross-agent tools:

ToolClaudeCodexAmpKimiOpenCodeGrokRoster score
SpecStory CLI❌ (open request #146)2/6
Mantra1/6
claude-replay3/6
coding_agent_session_search[UNVERIFIED — provider list not fully published]✅ likely~3/6
vibe-kanban (captive)4/6, not exportable
Nobody0 tools cover Kimi or Grok

Two consequences: (a) no off-the-shelf tool solves jackin's roster — best case 3/6; (b) Kimi and Grok providers would be first-in-market, and both turn out to have good native surfaces (Kimi versioned wire.jsonl + ACP; Grok documented session dirs + ACP updates.jsonl10).

Mechanism scorecard (matches Déjà Vu's vector taxonomy)

MechanismUsed byVerdict from field evidence
A. Native-store readSpecStory CLI, Mantra, all OSS exportersWins everywhere it exists: lossless, passive, post-hoc capable. Risk = format churn; mitigated by fingerprinted disposable indexes and vendor-blessed hedges
B. HooksLangfuse route, disler's observability, claude_telemetryGood liveness signal; only Claude hands back transcript_path; nobody uses hooks as sole capture
C. OTelGrafana official, claude-code-otel, LogfireMetrics/cost only; content redacted by design; never a transcript source
D. Wire tapclaude-trace, LeanMCP proxyOnly source of system prompts + guaranteed thinking; operationally invasive; niche
E. Vendor stream/APIAmp --stream-json, OpenCode serve, Codex --jsonThe correct answer where no file exists (Amp) and the hedge where formats are declared unstable (Codex)
F. Screen scrapenobody in the entire marketConfirms Déjà Vu's last-resort ranking — no shipping product relies on it

Tier list

TierToolRationale (one line each)
SSpecStory CLIOnly production-grade cross-agent capture pipeline with reusable Apache-2.0 source; 2/6 roster coverage but the mechanism blueprint (SPI, schema, index, reconstruct) transfers wholesale
AMantraProves the turn↔git-state layer users want next; closed source caps it at "study the UX, not the code"
Aclaude-replayBest output artifact in the market (portable HTML replay), 3/6 roster, MIT
Acoding_agent_session_searchBest retrieval breadth (11+ providers), same-day maintenance; pairs with any capture layer
Aclaude-code-logThe Claude Code archival workhorse; parse rules worth mirroring
Bclaude-code-transcriptsBest publishing/sharing workflow (gist, SSO sites); single-agent, slowing
Baichat search / claude-history / claude-code-history-viewerHealthy single-purpose retrieval/browsing tools; validate index-over-store design
BAmp threads · opencode /shareNative benchmarks any cross-agent surface must match (and interop with, since jackin runs both agents)
BLangfuse hook routeOnly observability platform that lands full coding-agent conversations; team features for free; Claude-only ingestion today
Cclaude-trace · LeanMCPWire-depth when system prompts/thinking must be exact; too invasive as default
Cccusage / usage monitors / viberank / vibe-logDifferent job (metrics); proves multi-product value of native stores
CHappy · vibe-kanban · Crystal · ConductorOrchestrator-captive history; category evidence, not reusable components
Fcursor-chat-export, cclogviewer, sniffly, claude-code-otel, opcode, omnara, cui, claude-code-exporterStale, archived, or upstream-deleted — the graveyard that shows maintenance is the hard part

What the matrix says in one paragraph

The market has one production cross-agent capture pipeline (SpecStory, open, 2/6 roster fit), one closed product pointing at the next layer (Mantra's git time-travel), a healthy single-agent Claude ecosystem, thin-to-nonexistent everything else, and zero coverage of Kimi and Grok. Capture mechanism is a solved question (native stores + vendor-stream hedges); differentiation now lives in normalization quality, retrieval, sharing surfaces, and turn↔code attribution. For jackin that reads: nothing to buy, one blueprint to port, two first-mover providers to write, and a defensible position from container ownership that no market player shares (30).

On this page