Plannotator v0.25.0

Plannotator is an open-source platform designed to review and validate plans and code generated by AI coding agents.

The platform provides a centralized interface where developers can:

• Review AI-generated plans

• Annotate proposed actions

• Approve or reject tasks

• Inspect code changes

• Provide structured feedback

• Improve agent outputs

• Maintain human oversight

Rather than replacing developers, Plannotator aims to keep humans involved in critical decision-making during AI-assisted development.

As AI coding agents become increasingly capable, developers face a growing challenge: ensuring that AI-generated plans and code changes align with project requirements before they are executed. Plannotator addresses this problem by acting as a review layer between developers and AI coding agents.

Instead of generating code itself, Plannotator focuses on making AI-driven development more transparent and controllable. It allows teams to review, annotate, approve, reject, and refine agent-generated plans and code changes before they affect a project.

Download Plannotator v0.25.0 - Software Mirrors
Plannotator v0.25.0 for Windows plannotator-paste-win32-arm64.exe | 106.86 MB plannotator-paste-win32-x64.exe | 110.08 MB plannotator-win32-arm64.exe | 160.67 MB plannotator-win32-x64.exe | 163.89 MB
Plannotator v0.25.0 for macOS plannotator-darwin-arm64 | 112.47 MB plannotator-darwin-x64 | 116.88 MB plannotator-paste-darwin-arm64 | 58.24 MB plannotator-paste-darwin-x64 | 63.07 MB
Plannotator v0.25.0 for Linux plannotator-linux-arm64 | 147.98 MB plannotator-linux-x64 | 148.52 MB plannotator-paste-linux-arm64 | 94.17 MB plannotator-paste-linux-x64 | 94.7 MB
Plannotator v0.25.0 Release Notes: Follow @plannotator on X for updates Missed recent releases? v0.24.2: Annotate YAML/JSON/TOML config files, XDG data directory support, Codex model catalog update, Cursor sandbox escape hatch v0.24.1: Annotate accepts parent-relative ../ file paths v0.24.0: PR/MR artifact gallery, GitButler review support, port ranges, expanded comment editor, OpenCode + Pi fixes v0.23.1: Startup no longer hangs on large or slow directory trees, Ask AI input stays visible after long responses v0.23.0: Plan approval fix for Claude Code 2.1.199+, annotate mode version diff, binary-only --minimal install, reviews post without attribution v0.22.0: Git-status "All changes" default review view, Commits panel with per-commit diffs, Guided Review, Pi + GitHub Copilot CLI review engines v0.21.4: Markdown math rendering, PR Overview panel with annotatable description and comments, agent instructions in code review, media parsing fixes v0.21.3: File comments in code review, unified click-to-highlight comments, VS Code clipboard/keyboard bridge, Codex Ask AI on app-server transport, CLI subcommand help v0.21.2: Custom reviews as Agent Skills, Cursor + OpenCode review engines, whole-file/general findings, deleted-annotation fix, Codex Ask AI outside git repos v0.21.1: Annotate-last blank-page fix on multi-message sessions v0.21.0: Direct document editing in annotate mode, live git-status file tree, in-app agent terminal, open files in external apps, HTML renders as HTML What's New in v0.25.0 This is the largest community release Plannotator has had. Eighteen pull requests landed since v0.24.2, thirteen of them from community members, and eight authors made their first contribution. The release adds an optional Vim keyboard layer, lets approvals carry your notes to the agent, makes gated annotate sessions scriptable, persists Guided Reviews across sessions, and hardens memory use, file watching, and multi-extension behavior across the board. Vim keyboard controls for plan review and annotate Plan review and annotate mode now have an optional Vim-style keyboard layer. Turn it on in Settings, then move through the document with j/k by block, refine into inline targets and exact text with l/h and the text motions, select with v/V, and create comments, redlines, and quick labels without touching the mouse. An optional HUD shows a target reticle, live keypress feedback, and a complete ? key map while you learn. It works in raw HTML annotate sessions too. Everything is off by default. If keyboard-driven review is not how you work, close the one-time announcement and nothing changes. Authored by @backnotprop in #1127 Approve with Notes Approving a gated annotate session used to throw away any annotations you had made along the way. Approval was a bare "approved" and your notes vanished. Now approval can carry them: the agent receives your annotations together with the approval across Claude Code, OpenCode, and Pi, with a customizable approvedWithNotes prompt template for tuning the wording. The approve flow also validates stale saved edits instead of silently proceeding, and approve-with-notes anchors to the correct message in multi-message annotate-last sessions. Authored by @rNoz in #1092, closing #930 requested by @tuanddd Strict automation mode for gated annotate plannotator annotate --gate --json always exited 0 no matter what the reviewer decided, so scripts could not use it as a real approval gate. A new strict mode fixes that: --require-approval exits nonzero unless the reviewer approves (configuration errors exit with a distinct code so automation never mistakes a typo for a rejection), and --result-file publishes the decision atomically with a write-temp plus no-clobber rename, so readers never see a partial record. The decision always reaches stdout before file publication, so a failed publish cannot destroy a completed review. Existing non-strict behavior is byte-identical. Authored by @rNoz in #1091, closing #1090 Annotate watches your file, not your whole folder Keeping an open source document synchronized with disk used to subscribe to the file's entire parent directory, so unrelated sibling files, nested changes, and git metadata churn all triggered pointless reconciliation. The watcher now tracks exactly the open files, in both the Bun and Pi servers, and survives atomic editor saves and platform quirks that previously could crash the watch on Linux. Authored by @rNoz in #1089, closing #1088 Guided Reviews persist across sessions A successful Guided Review is now saved automatically under your data directory, keyed to the repository. The guide screen shows your previous guides with their reviewed-progress, lets you reopen them read-through with per-section reviewed state intact, and flags a guide whose commit no longer matches the head under review. Disable persistence with PLANNOTATOR_GUIDE_HISTORY=0 or "guideHistory": false in config.json. Authored by @backnotprop in #1115, closing #1112 requested by @alexanderkreidich Folder annotate sessions get per-file version diffs Single-file annotate has shown a "since last review" diff badge for a while. Folder sessions now get the same treatment per file: open a folder, pick a file you have annotated before, and a +N/-M badge shows what changed since your last visit, with rendered and raw diff views. Each file keeps its own version history, shared with single-file sessions of the same file, and the diff view dismisses itself when you switch to a file with no baseline. Authored by @BenNewman100 in #1105, closing #1104 Bounded memory for large untracked files An unignored build artifact or large binary in git status could drive the review server to gigabytes of memory, and the freshness poll re-read the file every few seconds. Untracked files over 5 MiB now render as binary additions without loading their contents, freshness checks use size and mtime metadata instead of reading files, untracked diff generation is capped at four concurrent git processes, and oversized content is refused by the file-content endpoint in both runtimes. Authored by @kcosr in #1118, addressing the memory failure reported by @digitalmaster in #940 Disable AI features for managed deployments PLANNOTATOR_AI=disabled turns off Ask AI, Review Agents, and Guided Review launches across every server. The AI runtime is never constructed, so no agent processes are spawned or left idle, which matters on shared multi-user machines. Review and annotate flows work normally, saved guides remain readable, and capability probes degrade gracefully for older clients. Authored by @kcosr in #1129 OpenCode feedback goes to your agent, not a hardcoded one Review feedback in OpenCode was hardwired to hand off to an agent named build. If you had disabled or renamed it, every feedback send failed silently. Feedback now defaults to whatever agent you were already talking to, and a configured agent name is validated against your live agent list before sending, with a warning toast and graceful fallback when it is missing. Plan approval keeps its automatic build handoff, so the plan-to-build workflow is unchanged. Authored by @cb-bradbeebe in #1131, closing #1123 Pi extension plays well with other extensions Plannotator's Pi extension used to restore a snapshot of the global tool list when switching phases, wiping out tools other Pi extensions had activated or deactivated in the meantime. It now tracks exactly the tools it added for a phase and releases only those, leaving everyone else's selections alone. The planning phase always includes plannotator_submit_plan, even with a custom planning.activeTools configuration. Authored by @CodeByPeete in #1124 External plan execution handoff for Pi An opt-in executionMode: "external" setting lets a companion extension own execution. On approval, instead of entering the executing phase, Plannotator restores the pre-planning state and emits a plannotator:plan-approved event carrying the plan path, content, working directory, and any approval feedback. Automatic execution remains the default and is unchanged. Authored by @cgngtr in #1126 Faster Pi extension loading The Pi extension's imports now use exact .ts relative specifiers instead of extension probing, which cuts the module resolution work Pi's loader does on every session start. Authored by @dca123 in #1106 Additional Changes Zed Preview in Open In App - Zed's beta channel installs as a separate app on macOS; it is now its own entry in the Open In App menu, shown only when installed. On Linux and Windows the preview channel ships the plain zed binary, so the existing entry already covers it. By @backnotprop in #1130, closing #1119 requested by @graemefolk YAML block scalars in frontmatter - frontmatter values using | and > block scalars now parse correctly, including CRLF sources. By @ruaridhw in #1101 Copying file annotations uses the right template - the annotate-mode Copy button produced a plan-rejection message instead of annotate feedback. By @backnotprop in #1109, closing #1107 reported by @ZBQtesla Windows cross-drive path fix - annotation paths no longer break when the repository and temp directory live on different drives. By @BenNewman100 in #1117 Save is the first tab stop in the comment popover - keyboard users tab from the comment textarea straight to Save, with the visual layout unchanged. By @balasivagn in #1122, closing #1121 VS Code extension on Open VSX - the extension (v0.16.8) is published to Open VSX for VSCodium and other non-Microsoft editors, requested by @7tg in #1110 Pi npm packaging fix - the published Pi package now ships every module it imports; a missing file would have broken npm installs of this release GitButler review workflow docs - the README documents reviewing GitButler workspaces Install / Update macOS / Linux: bash curl -fsSL https://plannotator.ai/install.sh | bash Windows: powershell irm https://plannotator.ai/install.ps1 | iex Claude Code Plugin: Run /plugin in Claude Code, find plannotator, and click "Update now". OpenCode: Clear cache and restart: bash rm -rf ~/.bun/install/cache/@plannotator Then in opencode.json: json { "plugin": ["@plannotator/opencode@latest"] } Pi: Install or update the extension: bash pi install npm:@plannotator/pi-extension What's Changed fix(ui): parse YAML block scalars in frontmatter by @ruaridhw in #1101 perf(pi-extension): use exact .ts relative specifiers by @dca123 in #1106 fix(annotate): use annotate feedback template for clipboard copy by @backnotprop in #1109 feat: persist Guided Reviews across sessions by @backnotprop in #1115 fix(review): treat cross-drive relative() results as repo escapes by @BenNewman100 in #1117 fix(review): bound memory for large untracked files by @kcosr in #1118 feat: add an option to disable AI features by @kcosr in #1129 fix(ui): move Save to the first tab stop in the comment popover footer by @balasivagn in #1122 feat(core): add Zed Preview as an Open In App target by @backnotprop in #1130 feat(annotate): extend per-file version diff to folder sessions by @BenNewman100 in #1105 fix(annotate): watch open source files exactly by @rNoz in #1089 feat(annotate): add strict atomic result output by @rNoz in #1091 feat(annotate): preserve notes on structured approval by @rNoz in #1092 fix(opencode): fix hardcoded default build agent when sending responses by @cb-bradbeebe in #1131 feat(editor): add Vim keyboard annotation controls and live HUD by @backnotprop in #1127 fix(pi): preserve extension tool selections by @CodeByPeete in #1124 feat(pi): add external plan execution handoff by @cgngtr in #1126 docs: add GitButler review workflow by @backnotprop in #1111 New Contributors @ruaridhw made their first contribution in #1101 @dca123 made their first contribution in #1106 @BenNewman100 made their first contribution in #1117 @kcosr made their first contribution in #1118 @balasivagn made their first contribution in #1122 @cb-bradbeebe made their first contribution in #1131 @CodeByPeete made their first contribution in #1124 @cgngtr made their first contribution in #1126 Contributors @rNoz shipped a three-PR stack built from daily use: exact file watching, the strict automation gate, and Approve with Notes, stress-tested on a personal integration branch before ever reaching review. @BenNewman100 fixed Windows cross-drive paths and extended the version diff to folder sessions, keeping the branch current across three weeks of upstream churn. @kcosr contributed the memory bounds and the AI kill switch, both born from running Plannotator on shared agent infrastructure. @cb-bradbeebe diagnosed the hardcoded agent bug, proposed the fix on the issue, and delivered it with validation and tests. @CodeByPeete made the Pi extension a good citizen among other extensions, and @cgngtr built the external execution handoff on top of that model days after it landed. @ruaridhw fixed frontmatter block-scalar parsing, @dca123 made the Pi extension load faster, and @balasivagn improved comment-popover keyboard accessibility. Community reports and requests shaped the release throughout: @tuanddd requested approve-with-annotations back in #930 @digitalmaster reported the out-of-memory failure in #940 @alexanderkreidich requested Guided Review persistence in #1112 @ZBQtesla reported the clipboard template bug in #1107 @graemefolk requested Zed Preview support in #1119 @7tg requested Open VSX publishing in #1110 Full Changelog: https://github.com/backnotprop/plannotator/compare/v0.24.2...v0.25.0

Key Features of Plannotator

Plan Review System

One of Plannotator's core capabilities is reviewing plans generated by AI agents before execution.

Developers can examine:

• Proposed tasks

• Implementation strategies

• Agent reasoning

• Planned file modifications

• Workflow sequences

This visibility helps reduce unintended changes and costly mistakes.

Annotation Tools

The platform allows users to add comments, notes, and guidance directly to AI-generated plans.

These annotations can be used to:

• Clarify requirements

• Correct misunderstandings

• Provide context

• Guide future agent actions

AI Code Review

Plannotator extends the review process beyond planning by supporting inspection of generated code.

Developers can:

• Review modifications

• Analyze diffs

• Leave comments

• Request revisions

• Validate implementation details

This workflow resembles modern pull-request review systems.

Human-in-the-Loop Workflows

A major design goal is ensuring that AI actions remain subject to human approval.

Organizations can establish review processes where important actions require validation before execution.

Open Source Foundation

Plannotator is open source, allowing teams to inspect, modify, and self-host the platform according to their needs.

This transparency is particularly valuable for organizations adopting AI-assisted software development.

User Experience

The interface is designed around review workflows rather than direct code generation.

Instead of interacting with a chatbot, users primarily:

1. Receive agent-generated plans

2. Review proposed actions

3. Add feedback

4. Approve or reject changes

5. Monitor execution results

The workflow feels familiar to developers accustomed to pull requests, code reviews, and project planning tools.

Productivity Benefits

As AI coding tools become more autonomous, review processes become increasingly important.

Plannotator helps organizations:

• Reduce risky AI actions

• Improve code quality

• Increase accountability

• Preserve architectural consistency

• Encourage collaboration between developers and AI agents

For teams adopting AI-driven development, these safeguards can be as valuable as the coding agents themselves.

Collaboration Features

The platform supports collaborative review workflows where multiple team members can participate in evaluating AI-generated outputs.

This allows:

• Peer review

• Team approval processes

• Shared annotations

• Collective decision-making

Such features are especially useful for larger engineering teams.

Performance

Because Plannotator focuses on workflow management and review rather than model inference, performance largely depends on the connected AI agents and integrations.

The platform itself is lightweight and primarily serves as an orchestration and review layer.

Open Source Advantages

Being open source provides several benefits:

• Transparent development

• Self-hosting capabilities

• Custom integrations

• Community contributions

• Vendor independence

Organizations concerned about compliance, security, or proprietary workflows may find these advantages particularly appealing.

Limitations

Plannotator is designed as a companion tool rather than a complete AI development platform.

Common limitations include:

• Requires external AI coding agents

• Best suited for teams already using AI-assisted development

• Smaller ecosystem than mature developer platforms

• Additional review steps may slow rapid prototyping

• Some users may prefer fully autonomous workflows

The software delivers the most value in environments where oversight and quality control are priorities.

Pros

• Improves transparency of AI-generated plans

• Supports structured review workflows

• Human-in-the-loop design

• Useful annotation system

• Open source

• Self-hosting support

• Familiar review experience for developers

• Helps reduce AI-generated mistakes

Cons

• Not a standalone coding agent

• Requires integration with AI development tools

• Smaller community than established developer platforms

• Adds review overhead to workflows

• Best suited for teams rather than casual users

Who Should Use Plannotator?

Plannotator is ideal for:

• Software development teams

• Engineering managers

• AI-assisted development workflows

• Organizations adopting coding agents

• Open-source projects

• Teams prioritizing code quality and governance

It is particularly valuable for environments where AI-generated code requires oversight before reaching production systems.

Plannotator fills an increasingly important role in the AI development ecosystem by providing visibility and control over AI-generated plans and code changes. Its focus on human oversight, structured reviews, and collaborative workflows makes it a useful companion for modern coding agents. While it is not a replacement for AI coding tools themselves, it offers a practical solution for teams seeking greater confidence and accountability in AI-assisted software development.

Free

Download

Software Informations:
Developer:
backnotprop
Operating System:
Windows / macOS / Linux
Date Added:
2026-07-27T23:02:24.777Z
Categories:

Related to Plannotator v0.25.0:

Post a Comment/Report Broken Link: