SaaS Metrics
shepherd-agents/shepherd
A runtime substrate that turns an agent's execution into a reversible, Git-like trace, so meta-agents can observe, fork, replay, and revert any run. Couples agent and environments in a copy-on-write fork ~5x faster than docker commit, with ~95% KV-cache reuse on replay. Framework built for meta-agents to supervise, optimize, and train other agents
View Origin LinkProduct Positioning & Context
AI Executive Synthesis
Maintaining shepherd's performance and efficiency guarantees, particularly for its core VcsCore component which manages agent execution environments. The goal is to ensure that driver binding contract resolution, schema validation, and compilation occur only once per VcsCore lifetime, minimizing overhead for exec operations and upholding the intended caching invariant.
This issue exposes a performance regression in shepherd's VcsCore component, where the binding contract caching invariant is broken. Redundant `driver.describe()` calls, schema validation, and contract compilation on every `exec` introduce unnecessary overhead. While functionally correct, this regression directly impacts the efficiency of agent execution, particularly for drivers with non-trivial `describe()` operations. This undermines shepherd's promise of optimized runtime performance and efficient resource utilization, which are critical for meta-agents supervising and training other agents at scale. Addressing this caching flaw is essential to maintain the platform's performance integrity and developer trust.
A runtime substrate that turns an agent's execution into a reversible, Git-like trace, so meta-agents can observe, fork, replay, and revert any run. Couples agent and environments in a copy-on-write fork ~5x faster than docker commit, with ~95% KV-cache reuse on replay. Framework built for meta-agents to supervise, optimize, and train other agents
Related Ecosystem & Alternatives
Discover adjacent products, open-source repositories, and developer tools sharing similar technical architecture.
Deep-Dive FAQs
What is shepherd-agents/shepherd?
shepherd-agents/shepherd is analyzed by our AI as: Maintaining shepherd's performance and efficiency guarantees, particularly for its core VcsCore component which manages agent execution environments. The goal is to ensure that driver binding contract resolution, schema validation, and compilation occur only once per VcsCore lifetime, minimizing overhead for exec operations and upholding the intended caching invariant.. It focuses on This issue exposes a performance regression in shepherd's VcsCore component, where the binding contract caching invariant is broken. Redundant `dri...
Where did shepherd-agents/shepherd originate?
Data for shepherd-agents/shepherd was aggregated directly from the GitHub Open Source community ecosystem, representing raw developer and early-adopter sentiment.
When was shepherd-agents/shepherd publicly launched?
The initial public indexing or launch date for shepherd-agents/shepherd within our tracked developer communities was recorded on June 24, 2026.
How popular is shepherd-agents/shepherd?
shepherd-agents/shepherd has achieved measurable traction, logging over 998 traction score and facilitating 61 recorded discussions or engagements.
Which technical categories define shepherd-agents/shepherd?
Based on metadata extraction, shepherd-agents/shepherd is categorized under topics such as: mcts-rl, meta-agents, meta-optimization, runtime-supervision.
Are there active development issues for shepherd-agents/shepherd?
Yes, we are currently tracking open architectural debates and bug reports for this project on GitHub. There are currently 4 active high-priority issues logged recently.
What are some commercial alternatives to shepherd-agents/shepherd?
Our semantic intelligence engine identifies potential commercial alternatives in the SaaS space, such as Agent 37, which offers overlapping value propositions.
How does the creator describe shepherd-agents/shepherd?
The original author or development team describes the product as follows: "A runtime substrate that turns an agent's execution into a reversible, Git-like trace, so meta-agents can observe, fork, replay, and revert any run. Couples agent and environments in a copy-on-writ..."
Active Developer Issues (GitHub)
Logged: Jul 5, 2026
Logged: Jul 5, 2026
Logged: Jul 5, 2026
Logged: Jul 5, 2026
Community Voice & Feedback
**Datapoint: the CLI lane works on Linux at 0.2.0 with `claude` 2.1.201.**
Just ran the exact repro on WSL2 Ubuntu 24.04 (Landlock jail), Python 3.12, `shepherd-ai` 0.2.0 (fresh venv), `claude` CLI 2.1.201 signed in via OAuth (subscription, no env keys):
```
shepherd demo write claude-readme > agent_task.py
python agent_task.py # → run retained, README.md edited
```
`run show --json` → `status: retained`, `enforcement: syscall_jail / confined_process`, changeset holds the README edit. So this doesn't look like global CLI-contract drift at 2.1.201 — the divergence is somewhere in the macOS/auth path.
**Why the envelope looks empty (and where the real error is).** The refusal message is built as `(stderr+stdout)[-300:]` (`shepherd_dialect/providers.py`, `ClaudeHeadlessProvider`). When the headless CLI fails, it emits a JSON *result envelope* with the human-readable error in the `result` field — which sits near the **front** of the JSON. The last 300 chars are just the `usage…termin...
Just ran the exact repro on WSL2 Ubuntu 24.04 (Landlock jail), Python 3.12, `shepherd-ai` 0.2.0 (fresh venv), `claude` CLI 2.1.201 signed in via OAuth (subscription, no env keys):
```
shepherd demo write claude-readme > agent_task.py
python agent_task.py # → run retained, README.md edited
```
`run show --json` → `status: retained`, `enforcement: syscall_jail / confined_process`, changeset holds the README edit. So this doesn't look like global CLI-contract drift at 2.1.201 — the divergence is somewhere in the macOS/auth path.
**Why the envelope looks empty (and where the real error is).** The refusal message is built as `(stderr+stdout)[-300:]` (`shepherd_dialect/providers.py`, `ClaudeHeadlessProvider`). When the headless CLI fails, it emits a JSON *result envelope* with the human-readable error in the `result` field — which sits near the **front** of the JSON. The last 300 chars are just the `usage…termin...
Update: Your repro does work fine on my machine (*as in, it's not failing)! I'm continuing to investigate, but have run into an unrelated logging issue. I think this workaround may get you up and running in the meantime:
1. Run `claude setup-token` to get an OAuth token.
2. Make this available via `export CLAUDE_CODE_OAUTH_TOKEN=…`
3. Proceed with `python agent_task.py`, etc.
I hope that works! I'll be back to get to the bottom of this in a couple of hours either way – have an IRL commitment.
1. Run `claude setup-token` to get an OAuth token.
2. Make this available via `export CLAUDE_CODE_OAUTH_TOKEN=…`
3. Proceed with `python agent_task.py`, etc.
I hope that works! I'll be back to get to the bottom of this in a couple of hours either way – have an IRL commitment.
Thanks for the detailed writeup! That is odd – this should work, I've been using the CLI lane internally. Investigating now.
Amazing! Thanks for coming back so quickly and happy to discuss anytime really (based in CEST)!
Hello, thanks for writing! These are all very interesting questions:
1. We have some solutions cooking for full-VM – we'd like to support Firecracker, Modal, etc. Would love to bounce this around with you and see how well this works with your target use cases! Let me get back to you on this, ideally tomorrow, but def in the next few days.
2. Fascinating – I wonder whether it could make sense to pass around VM state as a type supported by our substrate. We have this in mind for our AI agent chat sessions, i.e. that one should be able to pass an immutable Claude session state into a `@task` as input, and receive one back as output. But I wonder whether this might inadvertently duplicate filesystem state. I'm quite interested in supporting this, though!
3. I'm tagging Simon on this point.
This is probably the best channel at this stage, but we might set up a Discord if things get busier!
1. We have some solutions cooking for full-VM – we'd like to support Firecracker, Modal, etc. Would love to bounce this around with you and see how well this works with your target use cases! Let me get back to you on this, ideally tomorrow, but def in the next few days.
2. Fascinating – I wonder whether it could make sense to pass around VM state as a type supported by our substrate. We have this in mind for our AI agent chat sessions, i.e. that one should be able to pass an immutable Claude session state into a `@task` as input, and receive one back as output. But I wonder whether this might inadvertently duplicate filesystem state. I'm quite interested in supporting this, though!
3. I'm tagging Simon on this point.
This is probably the best channel at this stage, but we might set up a Discord if things get busier!
Discovery Source
GitHub Open Source Aggregated via automated community intelligence tracking.
Tech Stack Dependencies
No direct open-source NPM package mentions detected in the product documentation.
Media Tractions & Mentions
No mainstream media stories specifically mentioning this product name have been intercepted yet.
Deep Research & Science
No direct peer-reviewed scientific literature matched with this product's architecture.