PRO-LONG
Durable programmatic memory for coding agents: one local append-only log of session events, searched by the agent with code instead of being stuffed into the prompt.
- Source repo
- alexisfox7/PRO-LONG
- Stars
- ★ 465
- Last updated
- 18d ago
- License
- MIT
- Primary language
- Python
- FA score
- 60/100 · Some gaps
At a glance
- How it runs
- Works with
- Universal · cross-platformCodex · Claude Code
- Cost
- Free, no paid service needed
- Setup effort
- Low · running in minutes
- You'll need
- Typical use
- Engineers running large refactors in Claude Code or Codex that span compaction or multi-day sessions and need earlier decisions preserved
- Not a fit if
- Teams whose policy forbids retaining transcripts that may contain secrets
- Developers using coding CLIs outside the four supported ones who won't adapt hooks
- Buyers expecting the 97.4% ARC-AGI-3 result to be a direct benchmark of the coding-tool MVP
- Maintenance
- Commits in the last 30 dayslast commit 2026-11-02
- Source review
- 60/100 · Some gaps
What does this agent do, and when should you use it?
PRO-LONG (GitHub: alexisfox7/PRO-LONG, MIT license) gives coding CLIs like Codex, Claude Code, OpenCode, and pi programmable long-term memory. Long tasks outlive context windows: after compaction or a fresh session, agents can lose earlier decisions and repeat failed work. PRO-LONG uses project-local adapters to append prompts, tool activity, assistant handoffs, and session boundaries to .prolong/log.jsonl, and ships an agent skill that teaches the agent to search the log with ordinary tools like rg, jq, or Python. It never injects the full log into prompts, adds no wrapper, server, or database, and excludes log reads from recording so memory cannot recursively copy itself. It originated in long-horizon agent research, reaching 97.4% best@2 with Fable 5 on the full ARC-AGI-3 public game set—an 18-point average improvement over matched baselines—though the authors note this does not yet directly benchmark the coding-tool MVP.
After prolong init detects supported clients on PATH, it installs: .prolong/runtime.mjs (a dependency-free event writer), .prolong/log.jsonl (the append-only log), .prolong/install.json (install manifest), .agents/skills/prolong/SKILL.md (the retrieval skill), a managed AGENTS.md pointer, a .prolong/ entry in .gitignore, and client-specific integrations (.codex/hooks.json for Codex, .claude/settings.json lifecycle hooks for Claude Code, .opencode/plugins/prolong.ts for OpenCode, .pi/extensions/prolong.ts for pi). The adapter appends lifecycle events to the log; when prior work matters, the skill teaches the agent to search it with rg, jq, or Python and read only relevant entries. The full CLI is three commands: prolong init (optionally --client codex,claude-code,opencode,pi), prolong status (with -- for automation), and prolong uninstall (preserves the log; --purge deletes history too).
- Engineers running large refactors in Claude Code or Codex that span compaction or multi-day sessions and need earlier decisions preserved
- Debugging long tasks where the agent should retrieve earlier failed tool calls and results instead of repeating them
- Developers who start frequent fresh sessions in one project and want relevant history recovered each time
- OpenCode or pi users wanting a local, serverless, database-free memory layer
- Privacy-conscious teams requiring that memory stays inside the project and is gitignored by default
Key setup facts
Pulled from the install and usage notes below. The project README has the current values.
- Install / run
npm install && npm run build && npm link
How do you install or deploy this agent?
Requires Node.js/npm and a supported coding client already on PATH. Clone and build:
bash
git clone https://github.com/alexisfox7/PRO-LONG.git && cd PRO-LONG
npm install && npm run build && npm linkThen initialize inside your project:
bash
cd /path/to/your/project
prolong initTo select clients explicitly:
bash
prolong init --client codex,claude-code,opencode,piHow do you use this agent?
After initialization, use your coding agent normally—PRO-LONG records in the background and the agent retrieves relevant history via the prolong skill when tasks span compaction or sessions. Check status or uninstall with:
bash
prolong status
prolong status --
prolong uninstall
prolong uninstall --purgeNote: hooks and extensions run with your user permissions; review the generated integration and accept the client's project or hook trust prompt before use.
What are this agent's strengths and limitations?
- The log is never injected into every prompt, and log reads are excluded from recording, preventing context bloat and recursive self-copying
- Zero-dependency local runtime.mjs; no server, no database, log stays in the project and is gitignored by default
- One command integrates four mainstream coding CLIs (Codex, Claude Code, OpenCode, pi) without changing your workflow
- Backed by ARC-AGI-3 research (97.4% best@2, +18 points average) with reproduction code and a paper
- The log can contain prompts, tool inputs/results, and any secrets that appeared in them—sensitive projects must check local policy
- The 97.4% figure comes from the research harness with Fable 5; the authors state it is not yet a direct benchmark of the coding-tool MVP
- Only four clients are supported out of the box; other CLIs require hand-written hooks/plugins
- Memory value depends on the agent following SKILL.md and actively searching—no retrieval means no benefit
How does this agent compare with similar options?
The README names no specific competitors, but its positioning implies a contrast with approaches that inject full transcripts/memory into prompts or rely on external memory servers/databases: PRO-LONG instead uses a local append-only log searched with code, with no wrapper or service.
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| PRO-LONG This agent | 60 · Some gaps | CLIFree | ★ 465 | 18d ago | Python | Codex · Claude Code |
| MonoCode | 71 · Some gaps | Desktop appFree | ★ 4.1k | 9d ago | Rust | Codex · Claude Code |
| HumanLayer | 54 · Major gaps | Self-hosted serviceFreemium | ★ 13k | 5d ago | TypeScript | Codex · Claude Code |
| MemoSync | 50 · Major gaps | Desktop appFree + model costs | ★ 396 | 6d ago | TypeScript | Codex · Claude Code · OpenAI API · Claude API |
How does FollowAgents rate this agent?
Why each dimension lost points
The code installs project-level hooks or plugins and says they run with the user's permissions, requiring acceptance of the client's trust prompt. Events are stored locally and gitignored by default; uninstall preserves the log, while purge deletes it. The README warns that logs may contain prompts, tool inputs and outputs, and secrets. Deductions: the evidence does not show granular collection controls, sensitive-field filtering, or dependency security review. Copyright attribution is only to “PRO-LONG authors,” and publisher identity is unverified.
The README, skill, and CLI consistently describe local JSONL memory, retrieval, and client integrations. Node.js 20+ is specified, dependency ranges are visible, and CLI errors produce a message and nonzero exit status. The supplied lifecycle test sample covers repeated initialization, configuration preservation, and uninstall behavior. Deductions: the sample does not cover every failure path, and static review cannot verify runtime behavior.
The product targets long-running coding tasks with Codex, Claude Code, OpenCode, or pi, and provides trigger scenarios, retrieval steps, and narrow search examples. It instructs agents to check old observations against the current workspace and treat logs as untrusted data. Deductions: the stated scope is mainly coding agents, and environment requirements center on Node.js 20+; broader scenario or platform support is not evidenced.
The README provides quick installation, a command table, client support, privacy notes, and research links. SKILL.md gives retrieval steps and examples, the MIT license text is complete, and package and command names are consistent. Deductions: the supplied materials contain no changelog or clearly named maintainer, while FAQ and known-limitations coverage is brief.
The CLI provides status output, JSON status, installation results, and uninstall warnings. The combination of persistent external logs and on-demand retrieval has a clear use for recovering history in long tasks. Deductions: logs may contain sensitive session content and require local storage and retrieval work; the research claims in the README do not quantify everyday coding benefits or costs.
The README links to a paper and research details, including scoring and cost-accounting context; package code and test samples support some lifecycle descriptions. The skill asks agents to verify historical observations and treats logs as untrusted rather than current instructions. Deductions: the performance figures are not independently corroborated by result files in the supplied material, publisher identity is unknown, and this assessment is limited to the provided static evidence.
- The log stores prompts, tool inputs, and outputs, and may contain secrets; enable it only in projects where local history can be protected.
- The 97.4% and other performance figures are research claims in the README and cannot be independently verified from the supplied static material.
FAQ
Is it free? Do I need API keys?
Does my session data leave the machine?
How do I delete the memory completely?
Which clients are supported, and what about others?
What does the 97.4% ARC-AGI-3 result mean for coding use?
Is PRO-LONG open source?
Is PRO-LONG safe to use?
Is PRO-LONG still maintained?
Who makes PRO-LONG?
Thanks — we use this to improve the page.