mirror of
https://github.com/wu736139669/hapi.git
synced 2026-08-05 06:24:37 +00:00
Junmo KimandGitHub
e45fde51e9
* refactor(claude): thread session's selected model into SDKToLogConverter
Adds an optional selectedModel field to the converter's context, wired
from session.getModel() in the launcher, so a later commit can seed the
turn-1 contextWindow estimate for presets whose system/init model
arrives without the "[1m]" suffix. No behavior change yet.
* fix(claude): key contextWindow cache by model to stop 1M/200k flicker
The remote launcher re-emits system/init on every turn for the same
converter instance. Its init-time estimate only checked whether the
model string ended in "[1m]", but current claude CLI versions strip
that suffix from system/init for some 1M presets (fable[1m] arrives as
"claude-fable-5"), so the estimate guessed 200k for them. The one
authoritative value is result.modelUsage[<model>].contextWindow, which
arrives after the heuristic has already injected 200k into that turn's
assistant message and then gets clobbered back to 200k by the very
next turn's init - producing the observed 200k<->1M oscillation in the
web status bar.
Cache the authoritative contextWindow per model id instead of a single
session-wide number, and only let system/init seed a heuristic guess
for a model that has no cached value yet, so a same-model re-init no
longer downgrades an already-learned value.
Two observed facts about the CLI's model ids drive the design:
system/init.model and the result.modelUsage keys always agree with
each other within a session (both bare for plain/fable[1m], both
suffixed for opus[1m]/sonnet[1m]), while each per-turn assistant
message reports its model bare and thus can't distinguish a 200k plain
preset from its 1M "[1m]" variant on tiers where they share a base id.
So the cache is keyed on the raw id (init/result agree, no
normalization) and assistant injection looks the value up via
resolvedModel (the last init id) rather than the lossy message.model.
Keying raw keeps a plain preset and its [1m] variant on distinct
entries; looking up via resolvedModel also means sidechain (Task
subagent) messages carry the main session window rather than the
subagent's own, since the web status bar picks the most recent usage
message without filtering sidechains and would otherwise flicker to
the subagent's smaller window while it runs.
For presets whose init model arrives bare even though they are 1M
(fable[1m]), the originally-selected preset - which preserves the
"[1m]" suffix - seeds the turn-1 estimate, kept live across mid-session
model switches via updateSelectedModel() (called from the launcher on
every turn) so it never goes stale.
* fix(web): recognize [1m] suffix on full Claude model ids in budget fallback
getContextBudgetTokens already special-cased "[1m]" for short preset
values (e.g. "opus[1m]") but fell through to the default 200k budget
for full model ids (e.g. "claude-opus-4-8[1m]"), which is what the CLI
now reports once context_window isn't available and this fallback is
consulted. Check the suffix on that branch too so it stays a correct
last-resort even without a session-provided context_window.
* refactor(web): merge duplicate Claude context-budget branches
isClaudeModelPreset(trimmedModel) and the startsWith('claude-') branch
below it had become byte-for-byte identical bodies after the [1m]
suffix check was added to both. Merge them into one condition; no
behavior change.
* fix(claude): distinguish fable from fable[1m] when the CLI reports both bare
The per-model contextWindow cache keyed on the raw system/init model id,
on the assumption that a 1M preset and its plain form always land on
distinct ids. That holds for opus[1m]/sonnet[1m] (the CLI reports the
"[1m]" suffix on their init and result ids) but not for fable: the CLI
reports both "fable" and "fable[1m]" with the same bare id
"claude-fable-5". So the "seed only if not already cached" guard would
skip re-seeding when switching fable[1m] -> fable, leaving the stale 1M
in place until fable's own result arrived - the same switch flicker this
change set out to remove, just for fable specifically.
Fold the selected preset's "[1m]" back into the cache key
(computeContextWindowKey): when the init model arrives bare but the
session selected an "[1m]" preset, key the entry as "<id>[1m]" so the 1M
and plain variants stay distinct; ids the CLI already suffixed are left
as-is. Seeding, lookups, and the current-model result entry all use this
resolved key. Subagent result entries (e.g. haiku) keep their own raw id
so the session's "[1m]" is never folded onto a model that isn't the
selected one.
hapi CLI
Run Claude Code, Codex, Cursor Agent, Gemini, or OpenCode sessions from your terminal and control them remotely through the hapi hub.
What it does
- Starts Claude Code sessions and registers them with hapi-hub.
- Starts Codex mode for OpenAI-based sessions.
- Starts Cursor Agent mode for Cursor CLI sessions.
- Starts Gemini mode via ACP (Anthropic Code Plugins).
- Starts OpenCode mode via ACP and its plugin hook system.
- Provides an MCP stdio bridge for external tools.
- Manages a background runner for long-running sessions.
- Includes diagnostics and auth helpers.
Typical flow
- Start the hub and set env vars (see ../hub/README.md).
- Set the same CLI_API_TOKEN on this machine or run
hapi auth login. - Run
hapito start a session. - Use the web app or Telegram Mini App to monitor and control.
Commands
Session commands
hapi- Start a Claude Code session (passes through Claude CLI flags). Seesrc/index.ts.hapi codex- Start Codex mode. Seesrc/codex/runCodex.ts.hapi codex resume <sessionId>- Resume existing Codex session.hapi cursor- Start Cursor Agent mode. Seesrc/cursor/runCursor.ts. Supportshapi cursor resume <chatId>,hapi cursor --continue,--mode plan|ask,--yolo,--model. Local and remote modes supported; remote usesagent -pwith stream-json.hapi gemini- Start Gemini mode via ACP. Seesrc/agent/runners/runAgentSession.ts. Note: Gemini runs in remote mode only; it waits for messages from the hub UI/Telegram.hapi opencode- Start OpenCode mode via ACP. Seesrc/opencode/runOpencode.ts. Note: OpenCode supports local and remote modes; local mode streams via OpenCode plugins.hapi resume [sessionId]- List resumable sessions for this machine or resume one locally.
Resume a remote session locally
hapi resume
hapi resume <session-id>
hapi resume lists resumable sessions for the current machine. hapi resume <session-id> hands off an active remote session and opens the same HAPI session in the local terminal.
Authentication
hapi auth status- Show authentication configuration and token source.hapi auth login- Interactively enter and save CLI_API_TOKEN.hapi auth logout- Clear saved credentials.
See src/commands/auth.ts.
Runner management
hapi runner start- Start runner as detached process.hapi runner stop- Stop runner gracefully.hapi runner status- Show runner diagnostics.hapi runner list- List active sessions managed by runner.hapi runner stop-session <sessionId>- Terminate specific session.hapi runner logs- Print path to latest runner log file.
Both start and start-sync accept repeatable --workspace-root <path> (or --workspace-root=<path>). When set:
- The web
/browsepage surfaces scoped file trees rooted at those paths. - The runner refuses
list-directoryandspawn-sessionrequests for paths outside the configured roots. ~and~/fooare expanded.
Omitting the flag keeps the legacy behavior: no scoping, no /browse feature.
See src/runner/run.ts.
Diagnostics
hapi doctor- Show full diagnostics (version, runner status, logs, processes).hapi doctor clean- Kill runaway HAPI processes.
See src/ui/doctor.ts.
Other
hapi mcp- Start MCP stdio bridge. Seesrc/codex/happyMcpStdioBridge.ts.hapi hub- Start the bundled hub (single binary workflow).hapi server- Alias forhapi hub.
Configuration
See src/configuration.ts for all options.
Required
CLI_API_TOKEN- Shared secret; must match the hub. Can be set via env or~/.hapi/settings.json(env wins).HAPI_API_URL- Hub base URL (default: http://localhost:3006).
Optional
HAPI_HOME- Config/data directory (default: ~/.hapi).HAPI_EXPERIMENTAL- Enable experimental features (true/1/yes).HAPI_EXTRA_HEADERS_JSON- JSON object of extra headers to send on CLI → hub requests, e.g.{"Cookie":"CF_Authorization=..."}.HAPI_CLAUDE_PATH- Path to a specificclaudeexecutable.HAPI_HTTP_MCP_URL- Default MCP target forhapi mcp.
Runner
HAPI_RUNNER_HEARTBEAT_INTERVAL- Heartbeat interval in ms (default: 60000).HAPI_RUNNER_HTTP_TIMEOUT- HTTP timeout for runner control in ms (default: 10000).
Worktree (set by runner)
HAPI_WORKTREE_BASE_PATH- Base repository path.HAPI_WORKTREE_BRANCH- Current branch name.HAPI_WORKTREE_NAME- Worktree name.HAPI_WORKTREE_PATH- Full worktree path.HAPI_WORKTREE_CREATED_AT- Creation timestamp (ms).
Storage
Data is stored in ~/.hapi/ (or $HAPI_HOME):
settings.json- User settings (machineId, token, onboarding flag). Seesrc/persistence.ts.runner.state.json- Runner state (pid, port, version, heartbeat).logs/- Log files.
Requirements
- Claude CLI installed and logged in (
claudeon PATH). - Cursor Agent CLI installed (
agenton PATH) forhapi cursor. Install:curl https://cursor.com/install -fsS | bash(macOS/Linux),irm 'https://cursor.com/install?win32=true' | iex(Windows). - OpenCode CLI installed (
opencodeon PATH). - Bun for building from source.
Build from source
From the repo root:
bun install
bun run build:cli
bun run build:cli:exe
For an all-in-one binary that also embeds the web app:
bun run build:single-exe
Source structure
src/api/- Bot communication (Socket.IO + REST).src/claude/- Claude Code integration.src/codex/- Codex mode integration.src/cursor/- Cursor Agent integration.src/agent/- Multi-agent support (Gemini via ACP).src/opencode/- OpenCode ACP + hook integration.src/runner/- Background service.src/commands/- CLI command handlers.src/ui/- User interface and diagnostics.src/modules/- Tool implementations (ripgrep, difftastic, git).
Related docs
../hub/README.md../web/README.md