@redis/agent-filesystem
BAgent Filesystem is now a workspace-first system for agents. Redis is still the
Install
agr install @redis/agent-filesystem --target claudeWrites 12 files into .claude/skills/, pinned to git-1020d588.
- .claude/skills/agent-filesystem/.dockerignore
- .claude/skills/agent-filesystem/.gitignore
- .claude/skills/agent-filesystem/AGENTS.md
- .claude/skills/agent-filesystem/Dockerfile
- .claude/skills/agent-filesystem/LICENSE
- .claude/skills/agent-filesystem/Makefile
- .claude/skills/agent-filesystem/README.md
- .claude/skills/agent-filesystem/SKILL.md
- .claude/skills/agent-filesystem/docker-compose.yml
- .claude/skills/agent-filesystem/example.afs.config.json
- .claude/skills/agent-filesystem/go.mod
- .claude/skills/agent-filesystem/go.sum
Document
Agent Filesystem
Agent Filesystem is now a workspace-first system for agents. Redis is still the canonical store, but the active user surfaces are:
afs mcpfor agent clientsafs ws mount <workspace> <directory>for a normal local working directoryafs config set --mode mountbefore mount for a live Redis-backed mount- explicit checkpoints via
afs cp ...
The old direct-command storage surface is retired and should not be used as the mental model for this repo.
Fast Start
make commands
./afs auth login
./afs ws create demo
./afs ws mount demo ~/demo
cd ~/demo
Most Useful Commands
./afs ws create <workspace>
./afs ws import <workspace> <directory>
./afs ws mount <workspace> <directory>
./afs ws unmount <workspace-or-directory>
./afs ws list
./afs ws fork <workspace> <new-workspace>
./afs cp create <workspace> <name>
./afs cp list <workspace>
./afs cp restore <workspace> <name>
./afs fs grep --workspace <workspace> "pattern"
./afs config set --mode sync
./afs config set --mode mount
./afs status
./afs mcp
Working Model
- Redis stores the live workspace state plus checkpoint history.
- Sync mode gives you a real local directory that is reconciled with Redis.
- Mount mode exposes the live workspace directly through NFS/FUSE.
- MCP talks to the same workspace model without requiring a local directory.
- File edits update the live workspace immediately; create checkpoints explicitly when you want a durable restore point.
Read Next
README.mdfor the current product story and setup flowdocs/README.mdfor the current documentation indexplans/README.mdfor active and archived planning artifactsdocs/internals/repo-walkthrough.mdfor the current tree layoutskills/agent-filesystem/SKILL.mdfor the installable agent skill
Trustgrade B
- passBody integrity
Whether the stored document is plausibly the kind of file the artifact declares, rather than something fetched by mistake.
- passType matchnot applicable to this artifact type
Whether the artifact is really the kind of thing its metadata claims it is.
- passFreshness
How long since the source repository was last pushed to.
- passPrompt injection
Scans the artifact's own text for instructions aimed at your agent rather than at you.
- warnLicensecopyleft/unknown — index-and-link only
Whether the source repository declares an SPDX license permissive enough to redistribute.
How the grade is calculated
Each check contributes 0 points when it passes, 1 when it warns, and 2 when it fails. The total maps to a letter:
- Aevery check passed
- Bone warning
- Ctwo warnings
- Dprompt injection or body integrity failed, or three warnings
- Fone of those failed, and something else is wrong
These are automated hygiene checks, not a security audit, and not a dependency or vulnerability scan. A grade of A means nothing was flagged — not that the artifact is safe.
Versions
git-1020d588b7192026-07-31