v1.3.0

1.3.0 (2026-04-21)

Features

• cron: cron test with streaming RPC and manual-run source tag (#68) (e801c4d)
• cron: daemon-backed cron ls with STATUS column and error tracking (#66) (60c4255)
• cron: suspend/resume CLI + format-preserving frontmatter writer (#65) (cac274e)
• cron: suspended frontmatter field and cron reload CLI (#64) (502de06)
• cron: timeout frontmatter field + TIMEOUT column + scheduled-run enforcement (#67) (47c34e8)
• memory: configurable model and compressed transcript for memory agent (c1c1a5d)
• memory: log memory agent runs and forbid in-session memory edits (5c56a6a)
• todos: session-scoped todo domain, tools, and /todos command (#71) (4d83bc2)
• todos: sticky TUI region + inline tool-call labels (#72) (544bc57)
• tui: cap tool-call lines at 4 per turn with overflow summary (121c29c)
• tui: styled todo region, insert tool calls above stream, preserve scroll (4a50700)

v1.2.0

1.2.0 (2026-04-12)

Features

• setup: interactive first-run setup wizard for LLM providers (#59) (c9b0cbe)

v1.1.0

1.1.0 (2026-04-11)

Features

• tool: add list_models tool to expose available models to the agent (eced02b)
• tool: add spawn_agent tool for ephemeral subagents (#49) (9f7a3d6), closes #40
• tui: surface subagent tool calls with agent label (#54) (627d47d), closes #53

v1.0.0

1.0.0 (2026-04-11)

Features

• abstract fs and start writing test (327c318)
• abstract session manager to have multiple frontend (4307208)
• add background install capability and an home dir (de8c06a)
• add bash tool (06fe178)
• add cronjobs (0592f5b)
• add debug (a436afa)
• add ESC to interrup ui (4aa1cdc)
• add inform_user tool for mid-loop progress updates (680570f)
• add init command, fetch_url tool, and Google Search grounding (84999a4)
• add makefile (941df82)
• add makefiles (b01a293)
• add memory to agent (7c2cb40)
• add OpenAI and Ollama provider support (#11) (50a7c7a)
• add OpenAI-compatible provider constructors for Regolo, Mistral, and Ollama (294bc7b)
• add parallel tool calling (1d952b8)
• add slash commands with autocomplete for TUI and Telegram (#23) (bc5be7f)
• add telegram (99d8f2e)
• add telegram user authentication with OTP approval (ae40b65)
• add tests for homer (d10f2d5)
• add tools (cf01aa6)
• cycle philosophical verbs in thinking spinner (#1) (7c51317)
• first commit, wellcome gonesis (6c5ace4)
• implement code mode with file-system tools and specialized prompt (ffab863)
• implement inform and tool call callbacks in telegram bridge (c8a36e7)
• implement skills (727818f)
• implement tools tests (ade5179)
• improve Agent.md (95c56e6)
• multi-provider config with lazy instantiation (#32) (1da98d1)
• rename project to wildgeku (a0ab15d)
• show toolcalling via TUI (abe955a)
• support per-session --model override for chat and code commands (25410bc)
• use Telegram typing indicator instead of "..." placeholder (#24) (b191302)

Bug Fixes

• agent: prevent nil panic in prompt building and remove stdout noise during tool execution (4263afa)
• handle ESC during tool calling (0b80255)
• linting issue (c434b81)
• pass full command string to LLM for skill slash commands (8f70188)
• propagate context to ToolExecutor and harden debug logger (dccd6d7)
• resolve all golangci-lint issues (99aadf9)
• resolve lint issues in telegram auth and bridge (55151ab)
• search (8f70afc)
• update TUI session ID after /clean to prevent stale session errors (382f265)
• verbs (3a0df2d)