|
Conderis • Reviewed knowledge for coding agents
Shared context for coding agents, with a review step. Decisions, tasks and project instructions reach every agent only after a maintainer approves them.
|
80 % |
|
Hem: work history for coding agents · hem.tools
Hem keeps a structured, erasable history of your coding-agent work (tasks, decisions, corrections, outcomes) over MCP. No transcripts. Nothing injected unles...
|
80 % |
|
MaguraIAbs — your AI remembers, and answers for it
One context for every coding session, and a verdict on each: what was claimed, checked against git and the running box.
|
80 % |
|
todox — working memory for developers and their agents
todox is a working memory for developers and their coding agents: projects, tasks, and the log that survives every session.
|
80 % |
|
Diffward — Review what your coding agent changed
Review agent-written code by hunk, see who changed what, and surface files you never asked for.
|
80 % |
|
Claude Agent Harness — docs, checks and hooks for coding agents
One command installs rules, a docs kit, checks and session hooks into your repository. Coding agents keep context, stay in their zones and stop overpaying to...
|
80 % |
|
Alike — The semantic web for your agents.
Alike is how your coding agents brief each other without going through you. It runs under Claude Code, Codex and Cursor, and starts every session with what y...
|
79 % |
|
Hivemind — Run agents in parallel. Break nothing.
Hivemind runs coding agents side by side, keeps each one in its lane, and stops bad changes before they hit your branch.
|
79 % |
|
Gloss — understand what your coding agent says, from the same session
Highlight any line your coding agent wrote — Claude Code, Codex or Gemini CLI — press ⌥ Space, and a note explains it beside the terminal using the matching...
|
79 % |
|
Hunch — A shared record for AI agents
Keep decisions, completed work, and commitments available to AI agents, with their sources. Git-native engineering memory and a self-hosted state server.
|
79 % |
|
Kodel - An editor for AI-driven development
Kodel keeps every repository you work in, each with its own Claude Code or Codex session, shell and editor, in one macOS window.
|
78 % |
|
Poggle | Shared memory for coding agents
Poggle is shared memory for coding agents. Agents log what they do as they do it, so the next one — different tool, different account, different machine — pi...
|
78 % |
|
Codaban — The reviewer who actually reads the code
Opinionated AI code review for Forgejo, GitHub, and (soon) GitLab. Three depths, one reviewer, no silent checkmarks. Closed beta.
|
78 % |
|
WitWiki: one wiki for every coding agent on your team, with a name on every edit
One wiki your coding agents read and write over MCP, whichever client they run in. Every change says who made it, and any version can be restored.
|
78 % |
|
Tiden · Intent verification for coding agents
Tiden verifies that your coding agents built what you actually asked for. It captures that intent as specs, versioned like code, and checks every change agai...
|
78 % |
|
RUNOUT — understand the code your agent shipped
Your coding agent opens more pull requests than you can read. RUNOUT turns each one into five sharp questions about the change — for you, and for the teammat...
|
78 % |
|
Workain — a control room for your coding agents
Run Claude Code and Codex CLI agents in parallel, from one browser tab, on your own subscriptions — with an honest status board that never guesses.
|
78 % |
|
flmnt — Your whole team. Every agent. One version of the truth.
flmnt keeps the decisions your team makes current across every session, person and AI tool, so nobody builds on the old plan.
|
78 % |
|
Review Assist: free AI code review MCP server for GitHub
Review AI-written code as fast as agents write it. Review Assist is a free, open-source MCP server that turns a coding agent's session into an Intent Documen...
|
78 % |
|
Daikonbun — where your AI agents work under review
Daikonbun is a self-hosted workspace where your AI coding agents work under review. Your own agent drives the board over MCP; every task is independently rev...
|
78 % |
Detected social media networks