v3.1.3 - Bug fix and code cleanup

What's Changed

• Fix: Skip message ID injection on empty assistant messages (#463)
• Refactor: Simplify hooks imports
• Refactor: Move manual trigger handling out of hooks

Full Changelog: v3.1.2...v3.1.3

v3.1.2 - improve compression reporting and XML tag cleanup

What's Changed

• Improved compress notifications with net savings and compression ratio details.
• Hardened hallucination stripping to remove stray dcp-prefixed XML tags more reliably.
• Refined token usage formatting helpers and aligned OpenCode SDK/plugin dependencies.

Full Changelog: v3.1.0...v3.1.2

v3.1.1 - Fix hallucination stripping for dcp-prefixed XML tags

What's Changed

• Fix: Broadened hallucination stripping to catch all dcp-prefixed XML tags instead of only specific ones

Full Changelog: v3.1.0...v3.1.1

v3.1.0 - Add experimental message-mode compression

What's Changed

• Add experimental compress.mode: "message", which lets DCP compress individual raw messages instead of only full ranges.
• Prioritize compression targets by message size, keep protected refs and completed compress calls visible, and preserve IDs across multipart content for more reliable message-mode behavior.
• Improve range mode with multi-range compression support and add compress.summaryBuffer so active summaries can extend the effective maxContextLimit in long sessions.
• Fix stale token and message-id handling after native /compact, count tool arguments more accurately, and clean up prompt/schema/config behavior.
• Add the scripts/opencode-message-token-counts inspector, expand compression and token-usage test coverage, and add GitHub issue templates.

Full Changelog: v3.0.4...v3.1.0

v3.0.4 - Bug fixes and config improvements

What's Changed

• config: Raise default compress thresholds for better context management
• compress: Ignore invalid block placeholders instead of erroring
• prompts: Fix override tag parsing

Full Changelog: v3.0.3...v3.0.4

v3.0.3 - Fix compress subagent refs and metadata tag leaking

What's Changed

• Fix compress tool to rebuild message refs when operating in subagent context
• Add instruction to prevent model from outputting injected XML metadata tags
• Add compress subagent integration test

Full Changelog: v3.0.2...v3.0.3

v3.0.2 - Fix host permissions compatibility

What's Changed

• Replace findLast with compatible alternative in host permissions for broader runtime support

Full Changelog: v3.0.1...v3.0.2

v3.0.1 - Prompt and permissions improvements

What's Changed

• Update context management guidance in compress prompts
• Align compress gating with opencode host permissions
• Standardize dcp-system-reminder XML naming convention
• Update demo readme images

Full Changelog: v3.0.0...v3.0.1

v3.0.0

v3.0.0 — Single Compress Tool Architecture

A major release featuring a complete architectural overhaul of the Dynamic Context Pruning system.

Breaking Changes

• The previous 3-tool system (distill, compress, prune) has been replaced with a single compress tool. All context management now flows through one unified interface.

New Features

• Decompress / recompress commands — Built-in commands for managing compression blocks
• Subagent support — experimental.allowSubAgents enables compression awareness in sub-agent contexts
• Custom prompts — experimental.customPrompts allows user-defined prompt overrides
• Flat schema option — compress.flatSchema for simplified tool schema presentation
• Protected user messages — protectUserMessages prevents user messages from being compressed
• Protected tools with glob patterns — Configure which tool outputs are preserved during compression using glob matching
• Visual progress bar — Real-time compression activity indicator
• Configurable turn nudges — Fine-tune when and how compression nudges appear
• Strengthened manual mode — More robust manual compression control

Improvements

• Reduced cache invalidation — Significantly decreases how often DCP causes cache invalidation, fixing Anthropic-related cache issues without needing manual mode
• Infinite conversations — Conversations can now last almost indefinitely; the previous tool pruning approach limited this by leaving user/AI messages untouched
• Simplified model behavior — The model no longer needs to choose between 3 context management tools, reducing decision complexity and improving reliability

Contributors

• @spoons-and-mirrors

Migration

This is a major version bump. Users upgrading from v2.x should review the updated README for the new single-tool configuration. The compress tool now handles all context management operations that were previously split across multiple tools.

Full Changelog: v2.1.8...v3.0.0

v2.2.9-beta0 - New config options and context compression improvements

What's Changed

• Added protectUserMessages config option to prevent user messages from being compressed.
• Added flatSchema config option for simplified tool schema injection.
• Added wildcard/glob pattern support for protectedTools configuration.
• Gated compress tool behind a one-shot manual trigger.
• Fixed: insert injected text parts before tool parts in assistant messages.
• Fixed: include cache.write in system prompt token calculation.
• Fixed: strip hallucinated tags from merged sub-agent results.
• Consolidated user and assistant turn nudges into a single prompt.
• Show raw LLM summary in showCompression notification.

Full Changelog: v2.2.8-beta0...v2.2.9-beta0