extensions
First-party extension manifests.
| Name | Description | Source |
|---|---|---|
agent_state # | Read-only introspection tool for the running agent | extensions/behaviors/companions/agent-state/manifest.fnl:1 |
builtin_tools # | Built-in fen workspace tools and on-demand tool discovery. | extensions/behaviors/kernel/builtin-tools/manifest.fnl:1 |
claude_memory # | Read-only, bounded Claude Code project memory index | extensions/behaviors/companions/claude-memory/manifest.fnl:1 |
compact # | Context compaction command and agent tool: summarize older messages and keep recent context. | extensions/behaviors/companions/compact/manifest.fnl:1 |
decide # | Experimental: opt-in advisory decisions from TypeSafe's Jev on OpenRouter or OpenAI Decisions API, used by compaction, handoff hints, and busy-line routing. | extensions/behaviors/kernel/decide/manifest.fnl:1 |
default_prompt # | First-party default system prompt policy | extensions/behaviors/kernel/default-prompt/manifest.fnl:1 |
docs # | In-agent runtime documentation browser | extensions/behaviors/kernel/docs/manifest.fnl:1 |
essentials # | Essential fen slash commands: help and model selection. | extensions/behaviors/kernel/essentials/manifest.fnl:1 |
extensions_inspector # | Extension inspector: /extensions, /reload-extension, and extension detail panel. | extensions/behaviors/inspectors/extensions/manifest.fnl:1 |
goal # | Goal companion: bounded autonomous objective execution. | extensions/behaviors/companions/goal/manifest.fnl:1 |
goal_headless # | Headless presenter for bounded goal runs | extensions/adapters/presenters/goal_headless/manifest.fnl:1 |
handoff # | Session handoff command: summarize current context and seed a fresh session. | extensions/behaviors/companions/handoff/manifest.fnl:1 |
json # | One-shot structured JSON presenter for --presenter json | extensions/adapters/presenters/json/manifest.fnl:1 |
mem # | Runtime memory diagnostics command | extensions/behaviors/companions/mem/manifest.fnl:1 |
plan # | Plan companion: draft, revise, inspect, and approve read-only execution plans. | extensions/behaviors/companions/plan/manifest.fnl:1 |
print # | One-shot stdout presenter for --print | extensions/adapters/presenters/print/manifest.fnl:1 |
profiler # | Opt-in Lua instruction-sampling profiler with Speedscope and folded flame-graph exports via /profile | extensions/behaviors/inspectors/profiler/manifest.fnl:1 |
profiler_tool # | Explicit opt-in model-facing profiler tool for contributors; not embedded in release manifests. | extensions/behaviors/inspectors/profiler-tool/manifest.fnl:1 |
prompt # | Prompt inspector: /prompt command and prompt-fragment panel. | extensions/behaviors/inspectors/prompt/manifest.fnl:1 |
provider_anthropic # | First-party Anthropic Messages provider. | extensions/adapters/providers/anthropic/manifest.fnl:1 |
provider_mock # | Deterministic, scriptable mock provider for tests, smoke runs, and offline dev. | extensions/adapters/providers/mock/manifest.fnl:1 |
provider_openai # | First-party OpenAI provider family (Chat Completions, Responses, Codex subscription, and Codex OAuth auth). | extensions/adapters/providers/openai/manifest.fnl:1 |
provider_openrouter # | First-party OpenRouter provider (curated OpenAI-Chat-Completions-compatible gateway models). | extensions/adapters/providers/openrouter/manifest.fnl:1 |
provider_sakana # | First-party Sakana AI provider (OpenAI-Responses-compatible Fugu models). | extensions/adapters/providers/sakana/manifest.fnl:1 |
provider_shared # | Shared transport helpers for first-party provider adapters (retry/backoff). Registers no user-facing surface. | extensions/adapters/providers/shared/manifest.fnl:1 |
queue # | Queue inspector: /queue, /cancel-all, and queue panel. | extensions/behaviors/inspectors/queue/manifest.fnl:1 |
rpc # | Live wire-protocol child presenter for --presenter rpc | extensions/adapters/presenters/rpc/manifest.fnl:1 |
session_jsonl # | JSONL session backend | extensions/adapters/session-backends/jsonl/manifest.fnl:1 |
sessions # | Conversation and session lifecycle commands plus an agent-callable hot-reload tool. | extensions/behaviors/actions/sessions/manifest.fnl:1 |
simplify # | Simplify companion: fan out subagent reviewers over the diff and apply quality cleanups. | extensions/behaviors/companions/simplify/manifest.fnl:1 |
skills # | Agent Skills discovery and prompt fragment | extensions/behaviors/companions/skills/manifest.fnl:1 |
status # | Runtime status inspector: /status command and status panel. | extensions/behaviors/inspectors/status/manifest.fnl:1 |
stdio # | Line-mode stdin/stdout presenter | extensions/adapters/presenters/stdio/manifest.fnl:1 |
steering # | Steering/follow-up input pipeline handler and queue service consumed by the agent loop and queue commands. | extensions/behaviors/kernel/steering/manifest.fnl:1 |
subagent # | Delegate a task to a child fen agent with isolated context | extensions/behaviors/companions/subagent/manifest.fnl:1 |
todo # | Structured todo list companion tool, panel, and status item | extensions/behaviors/companions/todo/manifest.fnl:1 |
tui # | First-party termbox2 presenter | extensions/adapters/presenters/tui/manifest.fnl:1 |
web # | First-party LuaSocket/SSE browser presenter | extensions/adapters/presenters/web/manifest.fnl:1 |