Added Letta Code to the list of coding agents with a brief description.

Add pi-builder to Agent infrastructure

Add CommandCodeAI's Command Code CLI agent - learns coding style
via taste-1 neuro-symbolic AI with project-specific taste profiles.

…ework

Add Agentic Engineering Framework to Agent infrastructure

…ges and intro blurb

…arp/Tabnine, improve contributing section

…h, GitClaw

ORCH is a CLI runtime for typed multi-agent teams that manages Claude Code,
Codex, and Cursor as a coordinated task queue. Key differentiators vs other
entries in this section: formal state machine (todo→in_progress→review→done),
auto-retry on agent failure, inter-agent messaging (orch msg send/broadcast),
and an Ink/React TUI dashboard. MIT license, TypeScript strict, 417 commits.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Add Untether — Telegram bridge for CLI coding agents

Add OneCLI — credential vault for AI agents

…badge

Add Parallel Code to Session managers & parallel runners

Add Coasts to agent infrastructure

Add VibePod - CLI for running AI coding agents in isolated containers

Add skill-optimizer to Agent infrastructure

Add agx — checkpoint-based agent orchestrator

Add gate4agent - Rust transport for CLI coding agents

Add Praman — CLI agents for SAP UI5/Fiori test automation

Add AgentLint

…infra-entry

Add EchoCoding — voice-enabled audio experience for AI coding agents

Add sage — pure bash agent orchestrator

Add Darce — ultralight multi-model CLI coding agent

Add LionClaw to OpenClaw ecosystem

Add lifeos-cli to Agent Infrastructure

Add Nex to Agent infrastructure

- scripts/update-stars.py fetches live GitHub stars, updates ⭐ badges,
  and re-sorts entries within each section by star count.
- .github/workflows/update-stars.yml runs it every Monday 08:00 UTC
  (and on manual dispatch) using the built-in GITHUB_TOKEN, then
  commits the diff.
- Ran once locally to refresh current counts and re-order sections.

Rust TUI coding agent from leonardcser/smelt (17⭐, MIT).

Closes #56. Credit to @rmarquis for the original PR; applied manually
because the star-refresh re-sort produced a merge conflict.

- Claw Code entry: instructkr/claw-code → ultraworkers/claw-code
  (the repo was renamed after its unlock; GitHub redirects the old URL,
  so the badge still resolved but the listing should point at canonical)
- Add HarnessLab/claw-code-agent (418⭐): Python-only Claude Code
  rewrite with zero external dependencies
- Add shareAI-lab/claw0 (2.3k⭐) to OpenClaw ecosystem: tutorial repo
  walking through building a claw-AI harness from scratch

Verified each repo for code quality before adding (real implementations,
sustained development, accurate descriptions).

- Octomind (Muvon/octomind, 50⭐): Rust agent runtime with tap registry,
  MCP, OAuth 2.1+PKCE, ACP. v0.25.0 with 1,243 commits. Closes #58.
  Credit @donhardman.
- openHarness (zhijiewong/openharness, 88⭐): Claude Code alternative
  with 42 tools, MCP, multi-provider. Closes #60. Credit @zhijiewong.
- CLITrigger (HyperAITeam/CLITrigger, 4⭐): Self-hosted web UI for
  parallel-worktree orchestration. Closes #65. Credit @OSgoodYZ.
- Relay (jcast90/relay, 2⭐): Local-first orchestrator running inside
  Claude/Codex CLI via MCP; multi-crate Rust+TS. Closes #59.
  Credit @jcast90.
- linear-cli (phnx-labs/linear-cli, 0⭐): Single-file Python Linear
  client designed as a subagent tool. Closes #61. Credit @muqsitnawaz.

Each entry placed in correct descending-stars position; all original
PRs would have conflicted with main due to the star-refresh re-sort.

Verified each repo for code quality before adding (real implementations,
sustained development, accurate descriptions).

- Aeon (aaronjmars/aeon, 263⭐): autonomous agent framework on GitHub
  Actions orchestrating Claude Code across 90+ skills with self-healing
  loop, quality scoring, persistent memory. Closes #69. Credit @aaronjmars.
- Crab Code (lingcoder/crab-code, 54⭐): Rust-native Claude Code
  alternative — 27 crates, 4,700+ tests, multi-provider, layered
  config. Closes #70. Credit @lingcoder.
- AgentPlane (basilisk-labs/agentplane, 41⭐): Git-native workflow
  control layer wrapping Claude Code/Codex/Cursor/Aider in an
  auditable task→plan→approve→implement→verify→finish flow.
  Closes #71. Credit @densmirnov.
- Not Human Search (unitedideas/nothumansearch, 0⭐): MCP server +
  REST API search engine for agentic-readiness signals (llms.txt,
  OpenAPI, MCP, ai-plugin); 8,000+ indexed sites. Closes #68.
  Credit @unitedideas.

Each entry placed in correct descending-stars position. Original PRs
would have conflicted with main due to ongoing star-refresh re-sorts.

Verified each repo for code quality before adding (multi-contributor
signals, sustained development, no slop indicators).

Session managers & parallel runners:
- agent-deck (asheshgoplani/agent-deck, 2.4k⭐): worktree-aware TUI
  session manager for multi-agent flows. Closes #83. Credit @asheshgoplani
  via @Jelloeater's PR.
- agent-of-empires (njbrake/agent-of-empires, 2.2k⭐): Rust TUI + web
  manager for multi-CLI-agent work over tmux + worktrees. Closes #83.
  Credit @njbrake via @Jelloeater's PR.
- ntm (Dicklesworthstone/ntm, 313⭐): Named Tmux Manager for
  coordinating multiple AI agents across panes. Closes #83. Credit
  @Dicklesworthstone via @Jelloeater's PR.

Orchestrators & autonomous loops:
- The Factory (akashgit/remote-factory, 24⭐): self-evolving meta-harness
  with auto-eval-discovery + keep/revert experiment loops; 8 contributors.
  Closes #74. Credit @akashgit.
- great_cto (avelikiy/great_cto, 16⭐): 34-agent SDLC orchestrator with
  compliance gates and multi-platform support. Closes #79. Credit @avelikiy.
- Galley (shinpr/galley, 1⭐): supervised-review Go runtime with retry/
  escalate loops + PR handoff. Closes #85. Credit @shinpr.

Agent infrastructure:
- agent-lsp (blackwell-systems/agent-lsp, 25⭐): MCP server providing
  LSP intelligence to CLI agents — fills a real gap in the list.
  (Original PR #75 was self-closed; re-adding with credit to
  @blackwell-systems.)
- Agent FM (agentfm-ai/agent-fm, 19⭐): macOS menubar narration app for
  Claude Code/Codex sessions. Closes #82. Credit @AniketDeshpandeGit.
- AgentTier (agenttier/agenttier, 18⭐): K8s-native sandbox runtime for
  AI agents, with Sandbox CRD + gVisor isolation. Closes #84.
  Credit @hardikvasa.

Each entry placed in correct descending-stars position; original PRs
would have conflicted with main due to the star-refresh re-sort.