# fidelios

> FideliOS CLI — orchestrate AI agent teams to run a business

- **URL**: https://www.freshcrate.ai/projects/fidelios
- **Author**: mvzemtsov
- **Category**: Frameworks
- **Latest version**: `v0.0.54` (2026-05-28)
- **License**: MIT
- **Source**: https://github.com/fideliosai/fidelios
- **Language**: TypeScript
- **GitHub**: 2 stars
- **Registry**: npm (`fidelios`)
- **Tags**: `agents`, `ai`, `cli`, `fidelios`, `npm`, `orchestration`

## Description

FideliOS CLI — orchestrate AI agent teams to run a business

## Recent releases

| Version | Date | Urgency | Changes |
| --- | --- | --- | --- |
| `v0.0.54` | 2026-05-28 | High | # v0.0.54  > Released: 2026-05-28  Bug fix for FID-64: agents (CMO on TRA-90) were silently reading zero new comments and stopping replies, even when comments were present.  ## Fix: comment cursor 500 silenced agent replies  [#105](https://github.com/fideliosai/fidelios/pull/105) — the `?after=` comment-cursor query bound a bare JavaScript `Date` into a raw Drizzle `sql` template. The postgres driver cannot bind a `Date` in an untyped param slot, producing a `Buffer.byteLength … Received an inst |
| `v0.0.53` | 2026-05-25 | High | # v0.0.53  > Released: 2026-05-25  Two rendering fixes for the symptom reported in FID-58: agents on the Codex adapter write comments and descriptions whose Markdown carries literal `\n` escape sequences instead of real newlines, so the UI showed the raw escapes instead of formatted text.  ## Escaped newlines in rendered comments  [#100](https://github.com/fideliosai/fidelios/pull/100) — comment bodies that contained literal `\n` (and HTML entities such as `&#x20;`) rendered the raw escape chara |
| `v0.0.52` | 2026-05-19 | High | # v0.0.52  > Released: 2026-05-19  Four changes — three adopted from a fresh comparison against the `paperclipai/paperclip` upstream (optimistic locking for agent instruction files, a configuration-diagnostic skill, a blocked-work inbox) and one new FideliOS capability: native per-company merge coordination.  ## Doc locking for instruction bundles  [#88](https://github.com/fideliosai/fidelios/pull/88) — concurrent edits to an agent's instruction bundle (`AGENTS.md`, `SOUL.md`, …) went through a |
| `v0.0.47` | 2026-05-17 | High | # v0.0.47  > Released: 2026-05-17  The in-app file viewer now finds files anywhere in the repo, opens them on the host, and links straight to the file on GitHub.  ## Fix: files outside the workspace cwd were "not found"  A workspace `cwd` can point at a *subdirectory* of a git repo (e.g. `…/TraitTune_v2/marketing`). The file viewer searched only inside that `cwd`, so a referenced file living in a sibling tree (`…/TraitTune_v2/docs/marketing/exhibits/COMPETITOR_SCAN.md`) reported "File not found" |
| `v0.0.42` | 2026-05-07 | High | # v0.0.42  > Released: 2026-05-07  Comment-routing reliability, Ollama adapter Phase 2, and a wave of UI fixes. Seven merged PRs on top of v0.0.41.  ## Critical: UI @-mentions now wake agents  [FID-44](https://github.com/fideliosai/fidelios/pull/61) — comments posted in the FideliOS UI containing `@AgentName` were silently failing to wake the targeted agent. Root cause was two-fold:  - `findMentionedAgents` regex broke on multi-word names ("Magic Leela") and missed HTML-entity-encoded handles - |
| `v0.0.40` | 2026-05-05 | High | # v0.0.40  > Released: 2026-05-06  Ten improvements across adapters, server, and UI — all shipped and validated on main.  ## New: Ollama in-process adapter (Phase 1)  FideliOS now has a native **ollama-local** adapter that runs entirely in-process via the Ollama npm SDK — no external CLI subprocess, no hermes gateway needed.  - Streaming chat over `ollama.chat()` with live token output - Auto-discovers local models via `/api/tags` and merges them with known cloud model list (Kimi K2.6, DeepSeek |
| `0.0.39` | 2026-04-21 | Low | Imported from npm (0.0.39) |
| `v0.0.39` | 2026-04-21 | High | # v0.0.39  > Released: 2026-04-21 > Telegram Gateway plugin: **Test Connection** and **Create Topics** buttons now activate after you save your bot token.  ## What's fixed  After saving a valid bot token and chat ID in the Telegram Gateway plugin settings, the custom settings page kept showing:  > Not configured — set botToken and chatId in plugin settings  and both **Test Connection** and **Create Topics** buttons stayed permanently disabled — no matter how many times you re-saved the configura |
| `v0.0.39` | 2026-04-21 | High | # v0.0.39  > Released: 2026-04-21 > Telegram Gateway plugin: **Test Connection** and **Create Topics** buttons now activate after you save your bot token.  ## What's fixed  After saving a valid bot token and chat ID in the Telegram Gateway plugin settings, the custom settings page kept showing:  > Not configured — set botToken and chatId in plugin settings  and both **Test Connection** and **Create Topics** buttons stayed permanently disabled — no matter how many times you re-saved the configura |
| `v0.0.39` | 2026-04-21 | High | # v0.0.39  > Released: 2026-04-21 > Telegram Gateway plugin: **Test Connection** and **Create Topics** buttons now activate after you save your bot token.  ## What's fixed  After saving a valid bot token and chat ID in the Telegram Gateway plugin settings, the custom settings page kept showing:  > Not configured — set botToken and chatId in plugin settings  and both **Test Connection** and **Create Topics** buttons stayed permanently disabled — no matter how many times you re-saved the configura |

## Citation

- HTML: https://www.freshcrate.ai/projects/fidelios
- Markdown: https://www.freshcrate.ai/projects/fidelios.md
- Dependencies JSON: https://www.freshcrate.ai/api/projects/fidelios/deps

_Generated by freshcrate.ai. Indexes npm releases for AI-agent ecosystem packages._
