By obra
Log key decisions and problem-solving approaches during coding sessions to preserve critical reasoning across context compaction in long-running AI interactions.
Status: Experimental. This plugin has an automated test suite but has barely been tested in real Claude Code sessions. Expect rough edges.
A Claude Code plugin that logs decisions and problem-solving approaches to disk so they survive context compaction.
When Claude Code compacts context, it loses the details of what was tried and decided during the session. This plugin provides MCP tools to record that information as it happens, and hooks that inject it back into context before and after compaction.
Decisions persist across the lifetime of the project — every session can search decisions from prior sessions.
Problems track the approaches tried during a session. When a problem is opened, every failed and successful approach is logged. This prevents retrying dead ends after compaction wipes the conversation history.
| Tool | Purpose |
|---|---|
log_decision | Record a decision with options considered and rationale |
search_decisions | Search project decisions across all sessions by keyword or tags |
open_problem | Start tracking approaches to a problem |
log_approach | Record a failed or successful approach to an open problem |
close_problem | Mark a problem as solved with a resolution summary |
list_problems | List problems in the current session, optionally filtered by status |
get_context | Reload all session state (decisions + problems) after compaction |
Data is stored as JSON files under ~/.claude/decision-log/<project-slug>/:
~/.claude/decision-log/
<project-slug>/
decisions.json # project-lifetime decisions
sessions/
<session-uuid>/
metadata.json # session info (cwd, timestamps)
problems.json # session-scoped problems + approaches
The project slug is a 12-char SHA-256 hash of the git remote URL (falling back to cwd if not a git repo).
claude --plugin-dir /path/to/cc-plugin-decision-log
Add the repository URL to your marketplace configuration. See Claude Code plugin docs for details.
cd server
npm install
npm run build # tsc
npm test # 35 tests across integration, hooks, edge cases, and E2E workflows
Admin access level
Server config contains admin-level keywords
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
npx claudepluginhub obra/cc-plugin-decision-logCore skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Semantic search for Claude Code conversations. Remember past discussions, decisions, and patterns.
Writing guidance based on William Strunk Jr.'s The Elements of Style (1918)
Experimental skills for Claude Code Superpowers - new techniques and tools under development
BETA: VERY LIGHTLY TESTED - Direct Chrome DevTools Protocol access via 'browsing' skill. Skill mode (17 CLI commands) + MCP mode (single use_browser tool). Zero dependencies, auto-starts Chrome.
Continuous memory for Claude Code. Extracts, summarizes, and compresses conversations into tiered daily logs. Claude remembers what you did yesterday.
Memory compression system for Claude Code - persist context across sessions
Persistent memory system for Claude Code with enforced context
Curated persistent memory for Claude Code. Write gate prevents bloat — only behavior-changing facts get saved. Tiered architecture: daily logs, structured registers, and auto-loaded working memory.
Claude Mind - Give Claude photographic memory in ONE portable file. Share, version, and transfer your Claude's brain.
Persistent long-term memory for Claude Code via MCP — captures coding decisions, bugfixes, and context across sessions. Hybrid FTS5 + TF-IDF search with episode batching. Single SQLite DB, no external services. A lighter, lower-cost alternative to claude-mem (episode batching + a smaller model; cost savings are an internal estimate, not a measured benchmark).