Local session indexing and cross-agent work continuity for AI coding assistants.
One local index over every AI coding agent on your machine. Search your entire coding history, read a project's cross-agent timeline, and hand work from one agent to another without losing the thread. Everything stays on your machine: no account, no cloud, no telemetry.
(The CLI command is voyager; the GitHub project is sessionFlow. This repo is
not DSH-only — the DSH plugin
is one client of it, and the core is a general-purpose local tool.)
| What it is | A local index plus a continuity layer over the session history your coding agents already write to disk. |
| Why use it | Your history is split across eight formats. sessionFlow makes it one thing you can search, summarize, export and continue — and every claim in a continuation bundle names the event it came from, so you can check it. |
| Install | pipx install "voyager[all] @ git+https://github.com/HarryHeYu/sessionFlow.git@v1.3.0" — details |
| Try it (2 min) | voyager demo — synthetic data, no agents needed — quick start · 5-minute walkthrough |
| Current release | v1.3.0 · Python ≥ 3.10 · Windows / macOS / Linux |
| Community | Listed in the third-party awesome-dsh-plugin directory — a listing, not a certification |
| Known limitation | Native resume is not verified end to end — see supported platforms |
| Maintaining it | Known issues, compatibility matrix, release checklist |
Continue work in a different agent. You started something in Codex; Claude
Code should finish it. voyager continue compiles the goal, the files touched,
the commands, the errors and where the work stopped into a context package the
next agent reads — instead of you re-explaining an afternoon.
One project, several sessions. A feature is spread over four sessions and
three agents. voyager merge A B C groups them into a WorkThread, and
voyager thread summarize <thread> gives you one brief across all of them —
who contributed what, and which items are still open.
Trace a claim back to the event. A handoff says "all 42 tests pass". Every
claim in a bundle names the event it came from, and voyager evidence verify
tells you whether that event is still the one the reference was made from. It
does not decide whether the claim is true — see below.
Your AI coding history is fragmented:
- Codex sessions (rollout JSONL)
- Claude Code conversations (project JSONL)
- DSH runs (zstd JSONL)
- ZCode / Cursor / Kiro / Grok / Antigravity (SQLite, VSCDB, JSON...)
sessionFlow creates one local index out of all of them — pure local, no accounts, no cloud, no telemetry.
Search. Understand. Continue. Switch agents without losing context.
Cross-agent continuity is built in: merge context across sessions and
continue in any agent (voyager merge / switch / continue) — see
docs/ROADMAP.md for the shipped scope and
docs/POST-1.0.md for what's next.
Morning Codex implements the SQLite migration
Afternoon Claude Code reviews the code
Evening DSH keeps debugging the edge case
sessionFlow:
find the previous reasoning -> voyager search "sqlite migration"
restore the full context -> voyager continue <thread>
switch agents mid-work -> voyager handoff claude --from codex
Nothing is uploaded anywhere: sessionFlow reads the session files your agents already wrote, normalizes them into one SQLite index you can search, resume from, hand off to another agent, or query straight from inside an agent over MCP / the DSH plugin.
Eight agents keep eight different formats; Voyager normalizes them into one SQLite index you can search, resume from, hand off to another agent, or query straight from inside an agent over MCP.
$ voyager list
ID PROV UPDATED MSG TOOL TITLE
a1b2c3d4-... zcode 2026-09-13 01:13 162 206 重构存储引擎的写入路径
9f8e7d6c-... codex 2026-07-06 22:52 76 168 修复多线程下载器的竞态条件
5e4d3c2b-... claude 2026-07-17 12:46 89 70 分析数据集结构并设计评测脚本
Use cases live in docs/use-cases.md — lost context, agent switching, engineering memory.
Every agent keeps its own history in its own format: Codex writes rollout JSONL, Claude Code writes project JSONL plus a file-version chain, ZCode uses SQLite, DSH compresses JSONL with zstd. You work across all of them — Voyager makes that history one thing you can query.
- Cross-agent timeline per repo — what did all your agents do to this project, and when?
- Full-text search everywhere — find the session where someone ran that one command or touched that one file (CJK substring search works).
- Real exports — human-readable Markdown, or lossless JSON with both the normalized and the raw events.
- Resume where you left off — Voyager knows each platform's resume command and runs it for you.
# isolated CLI install — no virtualenv juggling (recommended)
pipx install "voyager[all] @ git+https://github.com/HarryHeYu/sessionFlow.git@v1.3.0"
# or with pip (user-level); `python -m pip` on Windows, where `pip` may not be on PATH
python -m pip install "voyager[all] @ git+https://github.com/HarryHeYu/sessionFlow.git@v1.3.0"Voyager is not on PyPI. The voyager name there belongs to an unrelated
nearest-neighbour search library, so a bare pip install voyager installs the
wrong project — always use the @ git+… form above. Publishing under a final
name is tracked in CHANGELOG.md.
The @v1.3.0 pins a release tag, which is what makes a reinstall reproducible
and is what the DSH plugin's own error messages recommend. Drop the @v1.3.0
to follow main instead. Check the installed version with:
voyager --version # must print 1.3.0Upgrading from v1.2.0 (or v1.1.0) needs no migration. Your index is read as-is: no rows change, and the only difference is one index the new version adds for its own speed. Rolling back is safe because nothing was migrated — v1.1.0 reads the same database and ignores the extra index.
[all] = DSH support (zstandard) + MCP server (mcp). The two are
optional and only needed for those features:
pip install "voyager @ git+https://github.com/HarryHeYu/sessionFlow.git@v1.3.0" # core
pip install "voyager[dsh] @ git+https://github.com/HarryHeYu/sessionFlow.git@v1.3.0" # + DSH
pip install "voyager[mcp] @ git+https://github.com/HarryHeYu/sessionFlow.git@v1.3.0" # + MCP serverWorking on Voyager itself:
git clone https://github.com/HarryHeYu/sessionFlow && cd sessionFlow
pip install -e ".[all,dev]" # editable + extras + pytest
python -m pytest tests/ -q # 1303 passed, 6 skipped (measured at v1.3.0)Python ≥ 3.10. Windows / macOS / Linux. If voyager is not on your PATH,
run it as python -m voyager.cli.
A continuation bundle does not just summarise what an agent said — every tool observation and every assistant claim carries a reference to the event it came from:
- `pytest -q` # exit 1 · seq 2 · #cbd4d070
All 42 tests pass. · seq 3 · #c5009a39
voyager show prints the same pointer on every line, so you can go from a claim
to the message behind it. And you can check whether that pointer still lands on
the same content:
voyager evidence verify codex:01ab… 3 c5009a39It answers with one of five states and a matching exit code, so a script can branch on it:
| State | Exit | What it means |
|---|---|---|
MATCH |
0 | the event at that position is the one the reference was made from |
MISMATCH |
1 | something else is there now — the transcript changed under the reference |
NOT_FOUND |
2 | no event at that position (or no such session) |
UNVERIFIABLE |
3 | the reference carried no digest, so only the position could be checked |
INVALID_REFERENCE |
4 | the reference is malformed — e.g. a digest that is not 8 hex characters |
--json gives the same thing machine-readable. voyager evidence verify --help
does not list these states or codes; this table and the command's own output
do, and the walkthrough has one runnable example per state
on the synthetic demo index.
What MATCH means, and what it does not. It means the indexed event is the
one the reference was made from. It does not mean the claim is true — nothing
here decides that — and it is not a signature: the fingerprint is eight hex
characters, and anyone who rewrites a transcript can recompute it. The source
check compares mtime and size, which is not byte equality. Both limits are
stated in the command's own output, on every path.
voyager demo seeds a small synthetic index — four agents working one
"authentication flow" thread across a working day. No real agent, no real
data, nothing to install beyond sessionFlow itself:
voyager demo
voyager --db ~/.voyager/demo.db search "authentication" # hits 2 agents
voyager --db ~/.voyager/demo.db search "JWT refresh token"
voyager --db ~/.voyager/demo.db show codex:demo-auth-01 # compiled session--db is a global option: put it before the subcommand (dashboard and
evidence accept it only there).
The demo index (~/.voyager/demo.db) is separate from your real one — delete it
whenever you like. Two things in its output look like errors and are not: every
session is marked [source missing], because the demo's events exist only in the
index and no transcript files were written to disk (real retained history gets
the same label after a provider rotates its files — still searchable, not
resumable); and show prints a #seq/digest pointer on each line, which is the
reference Evidence Verify checks. What that looks like:
All screenshots are generated from the synthetic demo
(scripts/make_screenshots.py) — no real user data in this README.
Before the first scan. voyager scan opens the session files your agents
already wrote to disk (~/.codex, ~/.claude, …) and builds a local index at
~/.voyager/index.db. Nothing is uploaded and the provider files are opened
read-only — but that history can contain source code, prompts, command output and
file contents. The dashboard and every continuation package are generated from
that index, so they inherit it: treat dashboard.html, handoff-*.md and
~/.voyager/bundles/ as private and do not paste them into issues or chats.
Start with one agent rather than all of them:
voyager scan --platform codex # codex | claude | zcode | dsh, comma-separatedOnce at least one of your agents has session history on this machine:
voyager scan # discover + index every supported agent
voyager list # all sessions, newest first
voyager list --repo myproj # sessions for one repo
voyager show <id> # full message / tool-call timeline
voyager search "tensorboard"
voyager repo E:/code/myproj # cross-agent timeline for a repositoryvoyager scan walks every supported agent's local storage and builds the
index at ~/.voyager/index.db; it is incremental and only re-reads what
changed. To keep the index current with zero manual steps, run a watcher in
the background (OS autostart, or your own launcher script — voyager does not
install one for you), polling every 5 minutes:
voyager watch --interval 300 # re-scan every 5 minutes, forever
voyager export <id> --format md # or --format json (includes raw events)
voyager resume <id> # launches the native agent on that session
voyager handoff <id> --to codex # context package for another agent
voyager continue # one command to pick your latest work back up
voyager thread list # WorkThreads: task-centric session groups
voyager switch codex # switch the active thread to another agent
voyager skill install # teach other agents about voyager
voyager brief # 48h digest of what every agent is doing
voyager files <id> # files the session touched
voyager diff <id> # Claude sessions: rebuilt before/after diffs
voyager stats # index statisticsCross-agent handoff is work continuation, not session migration
(the other agent cannot inherit hidden tool state or cached reasoning).
voyager handoff <id> --to codex extracts the session into a Context
Package (goal, instructions, files touched, commands, errors, where the
work stopped) and shows the launch command; add --launch to start it.
The target is told to read the package file and continue — works for
claude, codex and grok; other targets get the file to paste.
Same-provider pickup still uses native resume (codex resume, …).
One command to continue: voyager continue picks your newest session
and does the right thing — native resume for codex/claude/dsh/grok,
automatic handoff package for the rest. It is a write, not a read: it refreshes
the index first (an incremental scan of your provider directories), takes the
WorkThread lease while it works, and produces a file — --no-launch prints the
command instead of starting it, and -o decides where the package goes.
voyager continue --repo myproj --launch goes straight back into a specific project; multi-session
synthesis (voyager merge A B C) groups the work into a WorkThread
and cross-agent switch is one command (voyager switch codex — lease
aware, see docs/ROADMAP.md).
switch, continue, handoff and merge are four spellings of one
engine (continuity.handoff_thread — see
docs/DECISIONS.md D14), so they cannot drift apart. When
the work sits in a WorkThread, all four take the single-writer lease, prefer
the native resume, record the pending attach that lets the next scan adopt
the target agent's new session, and warn — never stash — on a dirty tree.
handoff is the export spelling: --to names the agent that will read the
package, so it always compiles one. --steal takes over a live lease
explicitly; without it a second agent is refused by name.
Goal-conditioned & budgeted: add --goal "finish adapter tests" to
rank the evidence, and --budget compact|balanced|full|Nk to cap the
bundle size (estimate printed). Without --goal/--budget the output is
unchanged.
Transcript mode (opt-in, experimental): voyager switch codex --mode transcript writes a NEW native session containing a flattened
user/assistant transcript (tools/state dropped) so the target resumes
natively. Only codex/grok pass the resume gate today (claude timed out,
dsh unverified) — the default remains the Continuation Bundle. This is
work continuation, not session teleportation: hidden tool state and
provider runtime state never move.
Everyday flow — brief to see what's moving, export to read one
session in full (a 2,915-message DSH session → a 20 MB Markdown file),
continue or handoff to pick it back up. Full recipes in
docs/WORKFLOWS.md.
Session ids are matched by prefix; if a prefix is ambiguous Voyager lists
the candidates and exits. resume runs the native agent's own command
(e.g. codex resume <id>); providers without a CLI resume path say so
explicitly instead of pretending.
Core functionality: Complete and operational.
Runtime auto-trigger: Claude Code, Grok and Codex have native SessionStart hooks, and Voyager registers them. Claude Code firing its hook is live-verified as of 2026-09-24, Grok as of 2026-09-25, and Codex as of 2026-09-28: a normally launched Codex session in the same repo received the tiered-v1 context before its first turn and continued the current WorkThread on a bare 继续.
The startup_continuity() function correctly discovers WorkThreads, auto-attaches sessions, and compiles continuation context. What varies per provider is whether a session start can reach that function without the user doing anything.
Provider classification:
Codex is now on the native-hook path: voyager integrate install codex writes
~/.codex/hooks.json, and the fail-open handler emits the shared
hookSpecificOutput.additionalContext envelope with tiered-v1 context before
the first turn. Its handler, identity propagation, and idempotent lifecycle
are covered by integration tests, and a live provider firing is verified
(2026-09-28): hook delivery, native auto-attach, and zero-touch continuation.
Codex zero-touch continuity — live-verified 2026-09-28:
| Capability | Status |
|---|---|
| native SessionStart hook | PASS |
| tiered-v1 context delivery | PASS |
| zero-touch cross-agent continuity | PASS |
| native auto-attach | PASS |
Open Codex normally in the same repo and type nothing but 继续: it continues
the current WorkThread without calling voyager_startup, voyager_continue,
or any retrieval command.
Boundary: this is semantic continuity, not native transcript teleportation. The provider still starts a new native conversation; Voyager supplies the goal, the recent real work and the repository state — not a replay of the previous session's hidden tool state.
| Provider | Skill | MCP | Status | What it means |
|---|---|---|---|---|
| Claude | Y | R | H — hook registered |
Native SessionStart hook installed; the provider has been observed firing it live (2026-09-24). H is a static capability reading, not live evidence |
| Codex | Y | R | H — hook registered |
Native SessionStart hook installed; the provider has been observed firing it live (2026-09-28) and a bare 继续 continued the WorkThread with no Voyager command |
| Grok CLI | Y | N | H — hook registered |
Native SessionStart hook installed; the provider has been observed firing it live (2026-09-25). H is a static capability reading, not live evidence |
| DSH | Y | N | N — no mechanism |
Best effort |
Single source of truth. The table below is published from
voyager/capability_matrix.py, which is also whatvoyager doctorandvoyager integrate status --deepread. Each cell is a declared capability capped by what this machine has actually observed, so the docs cannot claim a live verification the code never earned. Runvoyager doctorto regenerate the machine's view of it.
voyager search <text> [--limit N] [--json]
[--provider codex,claude] [--repo SUBSTR]
[--since 7d | 2026-09-30] [--until ...]
[--kind user,assistant,tool_call] [--tool SUBSTR] [--file SUBSTR]
[--origin human,provider_bootstrap] [--human-only]
The text is always one quoted FTS5 phrase -- pytest -q, a:b and "unbalanced
are ordinary text to a human and operators to FTS5 -- so the query never has to be
escaped by hand. The filters are the structured half, and they only ever narrow
where to look: --human-only excludes turns a provider injected through the user
channel, --since 7d accepts relative days or a date, and combining them is a
conjunction. Chinese substring search works because the index is trigram-based.
voyager thread list|show|create|attach # the basics
voyager thread activity <thread> [--json] # who contributed what
voyager thread summarize <thread> [--json] # one brief across every agent
voyager thread checkpoint <create|list|show|update|export|restore>
voyager thread close|reopen|archive <thread> # explicit lifecycle
voyager thread stale [--days N] [--json] # active but idle
A brief is a derivation over the canonical WorkThread -- deterministic, offline,
and independent of any model: the authoritative fields come from the thread
itself, the recent turns are grouped per agent and read oldest-first, and open
items exist only because a checkpoint recorded them. reopen warns when it would
leave two active threads for one repository, because ambiguity is a choice the
user makes, never something a timestamp settles.
Nothing picks one for you. voyager continue and voyager switch print every
candidate and exit non-zero; voyager status reports AMBIGUOUS; the bridge
ops answer with thread_ambiguous: true and a thread_candidates list, and the
one they return is labelled a default rather than a decision. Name the thread
to proceed:
voyager continue --thread thr_xxxx # or `voyager switch claude --thread ...`
voyager thread close <id> # if one line of work is finished
Recency may suggest a default; it never establishes intent. A session that was merely re-scanned must not be able to take the headline away from the session the work was handed to, which is also why the bundle names the rule that chose it.
voyager dashboard [--out PATH] [--repo R] [--json]
Renders one self-contained HTML file (default ~/.voyager/dashboard.html):
projects, WorkThreads, the active thread and its agent members, recent activity
with a client-side filter, checkpoints, provider health and the open items. It is
an observation panel, not a chat client -- no server, no CDN, no network requests,
no JavaScript dependencies. Everything it shows comes from the same sources as
voyager doctor and voyager thread summarize, so the page cannot tell a
different story from the commands.
voyager doctor [--json] # is this installation healthy, and why not
voyager verify [provider] [--json] # declared / observed / effective per provider
voyager integrate status --deep # every capability dimension, with its evidence
voyager db check [--json] # read-only integrity diagnosis
voyager db backup [--json] # consistent snapshot via SQLite's backup API
voyager db repair [--json] # plan by default; --apply runs the safe steps
voyager db compact [--json] # VACUUM (maintenance, deliberately not "repair")
db repair never runs VACUUM — that is db compact. db repair plans by default and
--apply is the authorisation; there is no confirmation bypass.
Verification levels — these are deliberately not the same claim:
| Level | Meaning |
|---|---|
| SUPPORTED | Voyager can read the provider's session data |
| CONFIGURED | the provider's native startup hook is registered on this machine |
| UNIT_VERIFIED | the handler and envelope are covered by tests, and a real payload was driven through it end to end |
| LIVE_VERIFIED | the provider itself was observed firing the hook |
| ZERO_TOUCH_LIVE_VERIFIED | that, and a bare 继续 continued the WorkThread with no Voyager command |
| Provider | Startup surface | Level |
|---|---|---|
| Claude Code | native SessionStart |
ZERO_TOUCH_LIVE_VERIFIED (2026-09-24) |
| Grok CLI | native SessionStart |
ZERO_TOUCH_LIVE_VERIFIED (2026-09-25) |
| Codex | native SessionStart (~/.codex/hooks.json) |
ZERO_TOUCH_LIVE_VERIFIED (2026-09-28, session 01a0e7a2-6b1f-7011-ad65-103e17fb1094: the tiered-v1 developer message arrived before the first 继续, and that turn made zero voyager_* calls; the run's trace is in ~/.voyager/logs/codex-hooks.jsonl) |
| ZCode | native hooks (~/.zcode/cli/config.json, hooks.events.SessionStart) |
UNIT_VERIFIED — configured, and the handler was driven end to end with a real payload; a provider-fired run is still pending |
| Cursor | native hooks (~/.cursor/hooks.json, sessionStart) |
UNIT_VERIFIED — same |
| Kiro | native hooks (SessionStart / AgentSpawn, .kiro/hooks/*.json) |
UNIT_VERIFIED — handler written and driven end to end; hooks are project-scoped, so installing is a per-project choice |
| Antigravity | native hooks (PreInvocation, ~/.gemini/config/hooks.json) |
UNIT_VERIFIED — same, and injecting only on the first invocation |
| DSH | none found — profiles/plugins/ACP only | SUPPORTED — wrapper only |
A unit test is never reported as a live verification here: ZCode, Cursor, Kiro and Antigravity are configured and their handlers are exercised against real payloads, but until the provider has actually fired the hook on this machine they stay at UNIT_VERIFIED.
Legend: Y = installed, R = registered, N = unsupported, A = available/manual setup needed.
Startup status: Y = zero-touch verified live, H = native hook registered, A = startup-assisted, N = no hook. The letter is derived from static capability and configuration only.
Live verification is now a separate, persisted thing: voyager verify reports, per provider, the declared state (what the code supports), the observed state (what this machine has actually seen, derived from an append-only verification_events table) and the effective state (the weaker of the two). Observation can only lower a claim, never raise it above what the code supports, and voyager verify is strictly read-only — it cannot manufacture the evidence it reports. Run voyager doctor for the whole picture. Both Claude Code's trigger (2026-09-24) and Grok's (2026-09-25) have been observed live, and both still report H. Run voyager integrate status for the configuration answer on your machine.
Claude Code reads SessionStart hooks from ~/.claude/settings.json. Codex reads them from ~/.codex/hooks.json. Voyager writes the documented nested shape for each provider and preserves unrelated hooks:
{
"hooks": {
"SessionStart": [
{
"matcher": "startup",
"hooks": [
{"type": "command", "command": "\"<abs python>\" \"<abs>/claude_session_start.py\"", "timeout": 120}
]
}
]
}
}voyager integrate install claude writes this for you (it is additive — your own hooks are preserved, and the file is backed up first). The installed command uses absolute paths, so it does not depend on the interpreter being on PATH.
For Codex, voyager integrate install codex registers an absolute SessionStart command in ~/.codex/hooks.json. Codex passes its native session_id and cwd on stdin; Voyager responds with hookSpecificOutput.additionalContext using the shared tiered-v1 continuation format. The hook is fail-open and never blocks a Codex session if the index is unavailable.
Three things are verified. The handler is verified end-to-end: it emits a protocol-valid payload, caps the injected context at 9,000 UTF-16 code units, spills the full bundle to ~/.voyager/context/, and exits 0. The registration is verified: voyager integrate status reads the file back. And the provider firing the hook is verified too (2026-09-24) — the manual run below recorded a SessionStart whose session_id belongs to Claude Code itself, not the verifier's synthetic one:
claude --debug hooks --init-only # expect: Found 1 hook matchers in settingsThat line was observed on 2026-09-24, and the real session_id it carried then completed the whole chain: transcript discovered → indexed → pending row resolved → session attached to WorkThread thr_0854d50b88, with no Voyager command. What is still unproven is that the model actually read and used the injected context, and no provider prints Y — the CLI letter comes from static configuration, not from this observation. See docs/DOGFOOD.md for the evidence table.
startup_continuity()handles discovery, attach, staleness detection- Auto-attach works when conditions are safe (exact repo match, single thread)
- Context compilation reuses existing ranker+budget+continuation pipeline
- A compiled context bundle is cached in the store and reused across session starts (5-minute TTL, invalidated by git changes or thread membership changes)
- Ambiguity protection: explicit error if multiple active threads
- Auto-registration creates config files for Codex/Claude (
voyager integrate install <provider>), including the native Claude Code hook
Recommended workflows:
- Native hook (Claude Code):
voyager integrate install claude, then restart Claude Code. No further action — if the hook fires, context is injected before your first turn. - Explicit commands:
voyager switch <agent>orvoyager continue [id] - MCP-assisted: in-agent tool call
voyager_startup(provider="codex", cwd="$PWD") - Skill guidance: read
SKILL.mdin the agent's skill directory
A (startup-assisted) means: the runtime will not reach Voyager on its own, so you must invoke it via one of the workflows above. For Codex this no longer applies once voyager integrate install codex has registered the native SessionStart hook (live-verified 2026-09-28): the hook reaches Voyager before the first turn, so voyager_startup is not needed and should not be called — startup context is the hook's job, and the skill is retrieval-only.
voyager integrate install claude # skill + MCP + native SessionStart hook
voyager integrate install codex # skill + MCP (~/.codex/config.toml)
voyager integrate status # per-provider truth, including whether the hook is registered
voyager integrate remove claude # removes only Voyager's entries, leaves your hooks aloneSee docs/DOGFOOD.md for the detailed verification procedure.
Install the Skill file into known agent directories:
voyager skill install # installs SKILL.md at ~/.{agent}/skills/voyager/SKILL.mdThe Skill instructs agents when to use Voyager commands and when NOT to (never export full sessions).
To register Voyager as an MCP server so agents can query it with native tools:
voyager integrate install codex # skill + MCP (~/.codex/config.toml)
voyager integrate install claude # skill + MCP + native SessionStart hook (~/.claude/settings.json)
voyager integrate status # per-provider truth, incl. whether the hook is registered
voyager integrate remove <provider> # undo every step (skill + mcp + bootstrap + hook)These commands now auto-create config files if they don't exist - no manual setup needed for first-time installation. Re-running is idempotent, and install merges rather than overwrites: your own hooks and settings are preserved, and the file is backed up before it is rewritten.
For verification of actual zero-touch startup behavior, see docs/DOGFOOD.md and claude_continuity_verdict.md.
| Platform | Source | Messages | Tool calls | Shell exit | File diffs | Tokens | Resume |
|---|---|---|---|---|---|---|---|
| Codex (CLI/VSCode/Desktop) | rollout JSONL | ✅ | ✅ | ✅ | ❌ | ✅ | ✅ codex resume |
| Claude Code | project JSONL + file-history | ✅ | ✅ | ✅ | ✅ version chain | ✅ | ✅ claude --resume |
| ZCode | SQLite (~/.zcode/cli/db) |
✅ | ✅ | ✅ | ✅ usage tables | ❌ desktop only | |
| DSH | zstd JSONL (~/.dsh/sessions) |
✅ | ✅ | ❌ | ❌ | ❌ | ❌ --profile required |
| Grok CLI | chat_history.jsonl + summary.json |
✅ | ✅ | ❌ | ❌ | ❌ | ✅ grok -r |
| Cursor | state.vscdb (SQLite) |
✅ | ✅ | ❌ | ❌ IDE only | ||
| Kiro IDE | workspace-session JSON | ✅ | ❌ not persisted | ❌ | ❌ | ❌ | ❌ IDE only |
| Antigravity | conversation SQLite (protobuf) | ❌ | ❌ IDE only |
The Resume column names the command Voyager will run for that provider — it
says the command exists and is wired up, not that a resume has been observed
to work. No native resume has been exercised end to end on any provider, so
treat every ✅ there as NOT VERIFIED until the verification table above
carries a row saying otherwise. (The *_continuity_verdict.md files cover hook
injection, which is a different capability.)
Cursor and Antigravity adapters are marked experimental: Cursor reads its key-value store read-only and Antigravity decodes protobuf blobs heuristically (no public schema). Full per-field availability matrix and data-source paths for every tool are in docs/RECON.md.
DeepSeek Harness integration: https://github.com/HarryHeYu/dsh-sessionflow
dsh-sessionflow is a thin DSH
plugin that exposes search, current work, continuation and merge as six
sessionflow_* tools, so a DSH agent can pick up work from Codex, Claude Code,
Grok and the rest without leaving the harness. Its current release is plugin
v1.1.0, tested against core v1.3.0 over bridge schema 1 — the two repos
are versioned independently, and the schema is what actually binds them.
It is a client of this repo, not a second implementation: the core stays the
single source of truth, and the plugin talks to it over the stable JSON surface
(voyager integration-info --json, voyager api, voyager merge --json).
Adapters are the part of Voyager that breaks when a vendor ships a storage change, so every platform has a regression test against a synthetic fixture — no real session data, no agent installation needed:
tests/
├── fixtures/ # codex/claude/dsh/grok/kiro JSON+JSONL, zcode/cursor/antigravity SQL seeds
├── conftest.py # builds tmp trees (incl. zstd + SQLite) and repoints adapters at them
├── test_codex.py test_claude.py test_zcode.py test_dsh.py test_grok.py
├── test_cursor.py test_kiro.py test_antigravity.py test_adapters.py
├── test_store.py test_export.py test_handoff.py test_cli.py test_mcp.py
├── test_claude_session_start_hook.py # the SessionStart entrypoint: protocol, size cap, spill, logging
├── test_codex_session_start.py # Codex envelope, rollout-authoritative cwd
├── test_zcode_session_start.py # ZCode envelope, camelCase payload aliases
├── test_cursor_session_start.py # Cursor top-level additional_context, workspace_roots
├── test_kiro_antigravity_session_start.py # Kiro raw stdout, Antigravity injectSteps gating
├── test_hook_payload.py # the shared UTF-16 cap, both-ends cut, spill
├── test_provenance.py # origin classification and the NULL sentinel
├── test_l1_bands.py # STRONG/WEAK/UNKNOWN/BOOTSTRAP_ONLY and the scheduler
└── test_context_cache.py # the persisted context cache, incl. hostile input
python -m pytest tests/ -q # 1303 passed, 6 skipped at v1.3.0 — adapters, store, continuity, budget, leases, switch, skill, API, MCP, integration, evidence, provider hooks
python scripts/run_tests_core_only.py # core-only simulated: a subset that skips the DB-backed suitesThose numbers are from this checkout with no optional extras installed. The
collected count does not depend on the extras; the pass/skip split does. With
mcp, zstandard, Pillow and the dsh extra present, the dependency-gated
skips disappear and what is left is the data-dependent class:
tests/test_unicode_preservation.py reads the default local index and skips
when it holds no Chinese-titled session (:33) or no sessions at all (:103).
On Windows one more skips — test_zcode_unreadable.py:162, because chmod does
not deny reads there. Neither class is a platform gate.
CI (.github/workflows/test.yml) runs the suite
on Python 3.10–3.13 (Linux) and 3.10/3.13 (Windows — the adapters deal with
%APPDATA%, drive letters and backslashes), plus a core-only job proving the
CLI works with zero optional dependencies. The store tests cover the
"don't wreck my thousands of sessions" contract: repeated scans never
duplicate, changed sources are re-parsed, vanished sources are pruned.
Provider files are read-only. Adapters translate each platform's events into
one normalized model (Session / Event) while keeping the raw provider
event alongside — nothing is lost, giant blobs are truncated with a pointer
back to the source. Everything lands in a local SQLite index with FTS5
(trigram, so CJK substring search works). Scans are idempotent: sources are
tracked by (mtime, size) and re-parsed only when they change; sessions
whose source files vanish are pruned.
Details in docs/architecture.md, docs/DECISIONS.md, docs/API.md and docs/FAQ.md and everyday recipes in docs/WORKFLOWS.md.
The Continuity Engine core is complete (see docs/ROADMAP.md for the full close-out). The post-1.0 backlog — VS Code Context Composer UI, auto-clustering research, Claude/DSH transcript gates, scoped scan, fs-event watcher, PyPI/packaging polish — lives in docs/POST-1.0.md.
Phased plan, CLI sketches, and the issue list: docs/ROADMAP.md · 中文.
MIT — see LICENSE.