Get This Tool
Mycelium
Pricing
- Model
- Free
Summary
AI agents are fast at building — they are not fast at deciding. Mycelium is a Claude Code harness that forces an agent to run discovery and weigh evidence before writing a single line.
The core problem Mycelium targets is the agent that sprints from idea to pull request without asking why the feature exists, who asked for it, or whether the scope is right. The tool inserts a structured discovery phase — sequential questions, evidence gathering, a brief — that the agent must complete before code generation begins. The vendor describes it as earning the right to start. It runs as a Claude Code plugin and is MIT-licensed, so you install it, keep the source, and own the workflow. The wall appears when your project outgrows Claude Code itself: teams that move to a different agent runtime drop Mycelium entirely because it has no standalone API and no runtime outside that harness.
Bottom line: Reach for Mycelium when your Claude Code agent keeps building the wrong thing; reconsider when your architecture moves off Claude Code, because the plugin has nowhere else to go.
Community Performance Report Card
No community ratings yet. Be the first to rate this tool!
Community Benchmarks Community
Sign in to submit a benchmarkNo community benchmarks yet. Be the first to share a real-world data point.
Pros
Sign in to edit- Inserts a structured discovery phase before code generation begins, so the agent cannot skip from idea to pull request without accounting for purpose and audience — the exact pattern that produces features nobody asked for.
- Produces a written decision brief as part of the workflow, which means scope decisions are documented at the moment they are made rather than reconstructed from memory in a post-mortem.
- MIT-licensed with self-hosted deployment and no external API dependency, so you audit the full workflow, modify it, and run it without a vendor in the data path.
- Designed specifically for Claude Code's plugin system, so setup is a single install command rather than a custom integration build — teams already on Claude Code absorb it without changing their toolchain.
- Covers four distinct project types — software, courses, AI tools, service design — which means the discovery logic is not narrowly tuned to one domain and does not require reconfiguration when the project type changes.
Cons
Sign in to edit- The plugin runs exclusively inside Claude Code: if your team moves to a different agent runtime — Cursor, a custom LangChain setup, any other harness — Mycelium stops working and the discovery workflow has to be rebuilt from scratch in whatever replaces it. That is the condition under which teams drop it for a competitor or a homegrown prompt chain.
- There is no API surface, which means the discovery workflow cannot be triggered programmatically from a CI pipeline, a product management tool, or any external system. Teams that want automated pre-flight checks on tickets or specs have to invoke it manually every time.
- The project is early-stage with a small contributor base — community reports from the GitHub activity suggest the docs describe intended behavior more than battle-tested edge cases, so teams running it on complex multi-stakeholder briefs are effectively doing their own QA on the questioning logic.
Community Reviews
Sign in to write a reviewNo reviews yet. Be the first to share your experience.
About
- Platforms
- Claude Code
- API Available
- No
- Self-Hosted
- Yes
- Last Updated
- 2026-07-05T22:48:51.929Z
Best For
Who it's for
- Teams using Claude Code who want agents to validate ideas before coding
- Developers avoiding scope creep on weekend projects
- Product teams needing evidence-based decision records
What it does well
- Software development projects
- Course and content creation
- AI tool development
- Service design and validation
Integrations
Discussion Community
Sign in to commentNo discussion yet. Sign in to start the conversation.
Compare Mycelium
Spotted incorrect or missing data? Join our community of contributors.
Sign Up to ContributeCommunity Notes & Tips Community
Sign in to contributeBe the first to contribute. General notes, observations, gotchas, and tips from people who use this tool day-to-day.
Frequently Asked Questions
- Is Mycelium free?
- Yes — Mycelium is fully free to use. There is no paid tier.
- Is Mycelium open source?
- Yes. Mycelium is open source.
- Can I self-host Mycelium?
- Yes. Mycelium supports self-hosting on your own infrastructure.
- What platforms does Mycelium support?
- Mycelium is available on: Claude Code.
Hours Saved & ROI Stories Community
Sign in to contributeBe the first to contribute. Concrete time/cost savings, with context. e.g. "Cut my code review backlog from 4h to 45m per week."
Curated lists that include this category
Agents that skip the ‘why’ ship code nobody asked for. Mycelium wraps Claude Code with a discovery harness: before the agent touches a file, it runs through a sequence of structured questions — purpose, audience, evidence for demand — produces a decision brief, and only then proceeds to code. The workflow is designed to pause at the right moment rather than sprint past it. The vendor frames this as the agent earning the right to start, not just starting.
The differentiating mechanic is the sequenced questioning loop. Rather than accepting a prompt and generating immediately, the agent holds in a discovery state, asking for and weighing evidence across multiple exchanges. That brief becomes a record — useful later when you are debugging why a feature was scoped the way it was, or defending a decision to a stakeholder.
Mycelium is built for software projects, course and content creation, AI tool development, and service design validation. It fits best when a team is using Claude Code already and has felt the sting of scope creep on a weekend project or a sprint that delivered the wrong thing. It does not fit when the agent runtime is anything other than Claude Code: there is no standalone API and no abstraction layer that would let it run elsewhere. Teams that migrate off Claude Code lose access to the tool entirely and take the workflow logic with them only if they rebuild it manually.
The tool installs as a Claude Code plugin — the README provides direct install commands — and is released under the MIT license with no paid tiers. Self-hosting is the only deployment mode. There is no commercial API, no SaaS wrapper, and no managed version, which means the operational burden stays with the team running it.